AI Generator Remover

AI Generator Remover — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

• 

  Peanut App

  Peanut, a product of Peanut App Ltd. is an online community for women who are planning to become pregnant, women who are pregnant, women who have had children, and women who are experiencing menopause. Profiles of potential friends are displayed to users who can swipe up to show intent to connect. Users can also connect via discussion threads, groups, and live audio conversations. The app allows users to select their stage of life (trying to conceive, pregnancy, motherhood, or menopause), so as to meet women at a similar life stage, and to discover relevant content. Peanut was founded by Michelle Kennedy shortly after she left Bumble, a female-first dating app. She has described Peanut as, "the app she wishes she had when she first became a mother". == History == Peanut was initially launched in 2017 for mothers and pregnant women. The app focuses on helping users find others with shared interests, such as spoken languages, occupations, and hobbies. It also displays a woman's life stage, such as the age of her children, or the stage of pregnancy. In 2018, it launched a community discussion feature that intended to give women an "alternative to other social platforms". In 2019, it started to serve women who are trying to conceive. In April 2021, it integrated live audio, in response to the COVID-19 pandemic, and the restrictions around in-person socializing. in September 2021, it started to include women who are navigating perimenopause, menopause, and postmenopausal. Although it had initially catered for younger women navigating into new families, a large number of users had undergone surgically or chemically induced menopause due to medical conditions. In July 2021, Peanut launched an investment micro fund, Peanut StartHER, focused on investing in women-owned businesses, as well as other historically excluded founders. == Operation == The Peanut app is a social network exclusively for women, focusing on topics of pregnancy, motherhood, fertility, and menopause. It is available on iOS and Android devices. Users must prove their identity, in keeping with the primary function of in-app safety, and then they can create a profile to interact with other users. For pregnant users, the “Bump Buddies” feature helps connect them with other Peanut users who have a similar due date, which aimed to help expecting mothers combat loneliness during the COVID-19 pandemic. Peanut users also have the option to join “Groups” ‒ sub-sections of users focused on specific topics, including (but not limited to) location, life stage, pregnancy due date, and interests or hobbies. The live voice chat feature “Pods”, enables Peanut users to socialize without the pressure of photos or video chat. It offers features such as a muted audience of listeners who need to virtually raise their hand to speak, emoji reactions, and hosts who can moderate the conversations and invite people to speak.

  Read more →
• 

  Residuated Boolean algebra

  In mathematics, a residuated Boolean algebra is a residuated lattice whose lattice structure is that of a Boolean algebra. Examples include Boolean algebras with the monoid taken to be conjunction, the set of all formal languages over a given alphabet Σ {\displaystyle \Sigma } under concatenation, the set of all binary relations on a given set X {\displaystyle X} under relational composition, and more generally the power set of any equivalence relation, again under relational composition. The original application was to relation algebras as a finitely axiomatized generalization of the binary relation example, but there exist interesting examples of residuated Boolean algebras that are not relation algebras, such as the language example. == Definition == A residuated Boolean algebra is an algebraic structure ( L , ∧ , ∨ , ¬ , 0 , 1 , ∙ , I , / , ∖ ) {\displaystyle (L,\wedge ,\vee ,\neg ,0,1,\bullet ,\mathbf {I} ,/,\backslash )} such that An equivalent signature better suited to the relation algebra application is ( L , ∧ , ∨ , ¬ , 0 , 1 , ∙ , I , ▹ , ◃ ) {\displaystyle (L,\wedge ,\vee ,\neg ,0,1,\bullet ,\mathbf {I} ,\triangleright ,\triangleleft )} where the unary operations x ∖ {\displaystyle x\backslash } and x ▹ {\displaystyle x\triangleright } are intertranslatable in the manner of De Morgan's laws via x ∖ y = ¬ ( x ▹ ¬ y ) {\displaystyle x\backslash y=\neg (x\triangleright \neg y)} , x ▹ y = ¬ ( x ∖ ¬ y ) {\displaystyle x\triangleright y=\neg (x\backslash \neg y)} , and dually / y {\displaystyle /y} and ◃ y {\displaystyle \triangleleft y} as x / y = ¬ ( ¬ x ◃ y ) {\displaystyle x/y=\neg (\neg x\triangleleft y)} , x ◃ y = ¬ ( ¬ x / y ) {\displaystyle x\triangleleft y=\neg (\neg x/y)} , with the residuation axioms in the residuated lattice article reorganized accordingly (replacing z {\displaystyle z} by ¬ z {\displaystyle \neg z} ) to read ( x ▹ z ) ∧ y = 0 ⇔ ( x ∙ y ) ∧ z = 0 ⇔ ( z ◃ y ) ∧ x = 0 {\displaystyle (x\triangleright z)\wedge y=0\ \Leftrightarrow \ (x\bullet y)\wedge z=0\ \Leftrightarrow \ (z\triangleleft y)\wedge x=0} This De Morgan dual reformulation is motivated and discussed in more detail in the section below on conjugacy. Since residuated lattices and Boolean algebras are each definable with finitely many equations, so are residuated Boolean algebras, whence they form a finitely axiomatizable variety. == Examples == Any Boolean algebra, with the monoid multiplication ∙ {\displaystyle \bullet } taken to be conjunction and both residuals taken to be material implication x → y {\displaystyle x\to y} . Of the remaining 15 binary Boolean operations that might be considered in place of conjunction for the monoid multiplication, only five meet the monotonicity requirement, namely 0 , 1 , x , y {\displaystyle 0,1,x,y} and x ∨ y {\displaystyle x\vee y} . Setting y = z = 0 {\displaystyle y=z=0} in the residuation axiom y ≤ x ∖ z ⇔ x ∙ y ≤ z {\displaystyle y\leq x\backslash z\ \Leftrightarrow \ x\bullet y\leq z} , we have 0 ≤ x ∖ 0 ⇔ x ∙ 0 ≤ 0 {\displaystyle 0\leq x\backslash 0\ \Leftrightarrow \ x\bullet 0\leq 0} , which is falsified by taking x = 1 {\displaystyle x=1} when x ∙ y = 1 {\displaystyle x\bullet y=1} , x {\displaystyle x} , or x ∨ y {\displaystyle x\vee y} . The dual argument for z / y {\displaystyle z/y} rules out x ∙ y = y {\displaystyle x\bullet y=y} . This just leaves x ∙ y = 0 {\displaystyle x\bullet y=0} (a constant binary operation independent of x {\displaystyle x} and y {\displaystyle y} ), which satisfies almost all the axioms when the residuals are both taken to be the constant operation x / y = x ∖ y = 1 {\displaystyle x/y=x\backslash y=1} . The axiom it fails is x ∙ I = x = I ∙ x {\displaystyle x\bullet \mathbf {I} =x=\mathbf {I} \bullet x} , for want of a suitable value for I {\displaystyle \mathbf {I} } . Hence conjunction is the only binary Boolean operation making the monoid multiplication that of a residuated Boolean algebra. The power set 2 X 2 {\displaystyle 2^{X^{2}}} made a Boolean algebra as usual with ∩ {\displaystyle \cap } , ∪ {\displaystyle \cup } and complement relative to X 2 {\displaystyle X^{2}} , and made a monoid with relational composition. The monoid unit I {\displaystyle \mathbf {I} } is the identity relation { ( x , x ) | x ∈ X } {\displaystyle \{(x,x)|x\in X\}} . The right residual R ∖ S {\displaystyle R\backslash S} is defined by x ( R ∖ S ) y ⇔ ∀ z ∈ X , z R x ⇒ z S y {\displaystyle x(R\backslash S)y\ \Leftrightarrow \ \forall z\in X,zRx\Rightarrow zSy} . Dually the left residual S / R {\displaystyle S/R} is defined by y ( S / R ) x ⇔ ∀ z ∈ X , x R z ⇒ y S z {\displaystyle y(S/R)x\ \Leftrightarrow \ \forall z\in X,xRz\Rightarrow ySz} . The power set 2 Σ ∗ {\displaystyle 2^{\Sigma ^{}}} made a Boolean algebra as for Example 2, but with language concatenation for the monoid. Here the set Σ {\displaystyle \Sigma } is used as an alphabet while Σ ∗ {\displaystyle \Sigma ^{}} denotes the set of all finite (including empty) words over that alphabet. The concatenation L M {\displaystyle LM} of languages L {\displaystyle L} and M {\displaystyle M} consists of all words u v {\displaystyle uv} such that u ∈ L {\displaystyle u\in L} and v ∈ M {\displaystyle v\in M} . The monoid unit is the language { ε } {\displaystyle \{\varepsilon \}} consisting of just the empty word ε {\displaystyle \varepsilon } . The right residual M ∖ L {\displaystyle M\backslash L} consists of all words w {\displaystyle w} over Σ {\displaystyle \Sigma } such that M w ⊆ L {\displaystyle Mw\subseteq L} . The left residual L / M {\displaystyle L/M} is the same with w M {\displaystyle wM} in place of M w {\displaystyle Mw} . == Conjugacy == The De Morgan duals ▹ {\displaystyle \triangleright } and ◃ {\displaystyle \triangleleft } of residuation arise as follows. Among residuated lattices, Boolean algebras are special by virtue of having a complementation operation ¬ {\displaystyle \neg } . This permits an alternative expression of the three inequalities y ≤ x ∖ z ⇔ x ∙ y ≤ z ⇔ x ≤ z / y {\displaystyle y\leq x\backslash z\ \Leftrightarrow \ x\bullet y\leq z\ \Leftrightarrow \ x\leq z/y} in the axiomatization of the two residuals in terms of disjointness, via the equivalence x ≤ y ⇔ x ∧ ¬ y = 0 {\displaystyle x\leq y\ \Leftrightarrow \ x\wedge \neg y=0} . Abbreviating x ∧ y = 0 {\displaystyle x\wedge y=0} to x # y {\displaystyle x\#y} as the expression of their disjointness, and substituting ¬ z {\displaystyle \neg z} for z {\displaystyle z} in the axioms, they become with a little Boolean manipulation ¬ ( x ∖ ¬ z ) # y ⇔ x ∙ y # z ⇔ ¬ ( ¬ z / y ) # x {\displaystyle \neg (x\backslash \neg z)\#y\ \Leftrightarrow \ x\bullet y\#z\ \Leftrightarrow \ \neg (\neg z/y)\#x} Now ¬ ( x ∖ ¬ z ) {\displaystyle \neg (x\backslash \neg z)} is reminiscent of De Morgan duality, suggesting that x ∖ {\displaystyle x\backslash } be thought of as a unary operation f {\displaystyle f} , defined by f ( y ) = x ∖ y {\displaystyle f(y)=x\backslash y} , that has a De Morgan dual ¬ f ( ¬ y ) {\displaystyle \neg f(\neg y)} , analogous to ∀ x ϕ ( x ) = ¬ ∃ x ¬ ϕ ( x ) {\displaystyle \forall x\phi (x)=\neg \exists x\neg \phi (x)} . Denoting this dual operation as x ▹ {\displaystyle x\triangleright } , we define x ▹ z {\displaystyle x\triangleright z} as ¬ x ∖ ¬ z {\displaystyle \neg x\backslash \neg z} . Similarly we define another operation z ◃ y {\displaystyle z\triangleleft y} as ¬ ( ¬ z / y ) {\displaystyle \neg (\neg z/y)} . By analogy with x ∖ {\displaystyle x\backslash } as the residual operation associated with the operation x ∙ {\displaystyle x\bullet } , we refer to x ▹ {\displaystyle x\triangleright } as the conjugate operation, or simply conjugate, of x ∙ {\displaystyle x\bullet } . Likewise ◃ y {\displaystyle \triangleleft y} is the conjugate of ∙ y {\displaystyle \bullet y} . Unlike residuals, conjugacy is an equivalence relation between operations: if f {\displaystyle f} is the conjugate of g {\displaystyle g} then g {\displaystyle g} is also the conjugate of f {\displaystyle f} , i.e. the conjugate of the conjugate of f {\displaystyle f} is f {\displaystyle f} . Another advantage of conjugacy is that it becomes unnecessary to speak of right and left conjugates, that distinction now being inherited from the difference between x ∙ {\displaystyle x\bullet } and ∙ x {\displaystyle \bullet x} , which have as their respective conjugates x ▹ {\displaystyle x\triangleright } and ◃ x {\displaystyle \triangleleft x} . (But this advantage accrues also to residuals when x ∖ {\displaystyle x\backslash } is taken to be the residual operation to x ∙ {\displaystyle x\bullet } .) All this yields (along with the Boolean algebra and monoid axioms) the following equivalent axiomatization of a residuated Boolean algebra. y # x ▹ z ⇔ x ∙ y # z ⇔ x # z ◃ y {\displaystyle y\#x\triangleright z\ \Leftrightarrow \ x\bullet y\#z\ \Leftrightarrow \ x\#z\triangleleft y} With this signature it remains the case that this axiomatization can be expressed as

  Read more →
• 

  Trevor Paglen

  Trevor Paglen (born 1974) is an American artist, geographer, and author whose work covers mass surveillance and data collection. In 2016, Paglen won the Deutsche Börse Photography Foundation Prize and he has also won The Cultural Award from the German Society for Photography. In 2017, he was a recipient of a MacArthur Fellowship. On March 17, 2026, Paglen was awarded the 2026 LG Guggenheim Award (a collaboration between LG and Guggenheim New York). == Early life and education == Paglen earned a B.A. degree in religious studies in 1998 from the University of California at Berkeley, a M.F.A. degree in 2002 from the School of the Art Institute of Chicago, and a Ph.D. in Geography in 2008 from the University of California at Berkeley. While at UC Berkeley, Paglen lived in the Berkeley Student Cooperative, residing in Chateau, Fenwick, and Rochdale co-ops. == Work == Sean O'Hagan, writing in The Guardian in 2015, said that Paglen, whose "ongoing grand project [is] the murky world of global state surveillance and the ethics of drone warfare", "is one of the most conceptually adventurous political artists working today, and has collaborated with scientists and human rights activists on his always ambitious multimedia projects." His visual work such as his "Limit Telephotography" and "The Other Night Sky" series have received widespread attention for both his technical innovations and for his conceptual project that involves simultaneously making and negating documentary-style truth-claims. Paglen’s work relies on contemporary technology in two meaningful ways. Firstly, the views he photographs would be impossible to shoot without media tech, that includes the cameras, the microscopes, and even helicopters. But interestingly enough, the shots would not be possible if not for the existence of the subject. The contrasts between secrecy and revelation, evidence and abstraction distinguish Paglen's work. With that the artist presents not so much "evidence" as admonitions to awareness. He was an Eyebeam Commissioned Artist in 2007. In 2008 the Berkeley Art Museum devoted a comprehensive solo exhibition to his work. In the next year, Paglen took part in the Istanbul Biennial, and in 2010 he exhibited at the Vienna Secession. Autonomy Cube was a project by Paglen and Jacob Appelbaum that placed relays for the anonymous communication network Tor in traditional art museums. He contributed to the Oscar-winning documentary film Citizenfour (2014), directed by Laura Poitras. Paglen features in the nerd-culture documentary Traceroute (2016). Orbital Reflector was a reflective, mylar sculpture by Paglen intended to be the first "purely artistic" object in space. The temporary satellite, containing an inflatable mylar balloon with reflective surface, launched into space 3 December 2018. A mid-career survey in 2018–2019, Trevor Paglen: Sites Unseen, was a traveling exhibition shown at the Smithsonian American Art Museum in Washington DC and the Museum of Contemporary Art San Diego. In September 2020, Pace Gallery in London held an exhibition of Paglen's work, exploring "the weird, partial ways computers look back at us". His work is included in the permanent collections of the San Francisco Museum of Modern Art, the Columbus Museum of Art, and the Metropolitan Museum. === Experimental Geography === Paglen is credited with coining the term "Experimental Geography" to describe practices coupling experimental cultural production and art-making with ideas from critical human geography about the production of space, materialism, and praxis. The 2009 book Experimental Geography: Radical Approaches to Landscape, Cartography, and Urbanism is largely inspired by Paglen's work. == Publications == Paglen has published a number of books. Torture Taxi (2006) (co-authored with investigative journalist A. C. Thompson) was the first book to comprehensively describe the CIA's extraordinary rendition program. I Could Tell You But Then You Would Have to be Destroyed by Me (2007), is a look at the world of black projects through unit patches and memorabilia created for top-secret programs. Blank Spots on the Map: The Dark Geography of the Pentagon's Secret World (2009) is a broader look at secrecy in the United States. The Last Pictures (2012) is a collection of 100 images to be placed on permanent media and launched into space on EchoStar XVI, as a repository available for future civilizations (alien or human) to find. === Publications by Paglen === I Could Tell You But Then You Would Have to be Destroyed by Me. Brooklyn, NY: Melville House, 2007. ISBN 1-933633-32-8. Blank Spots on the Map: The Dark Geography of the Pentagon's Secret World. New York: Dutton, 2009. ISBN 9781101011492. Invisible: Covert Operations and Classified Landscapes, Photographs by Trevor Paglen. New York: Aperture, 2010. ISBN 9781597111300. With an essay by Rebecca Solnit. The Last Pictures. Oakland, CA: University of California, 2012. ISBN 9780520275003. Trevor Paglen. London: Phaidon, 2018. ISBN 0714873446. With essays by Laren Cornell, Julia Bryan-Wilson, Omar Kholeif. === Publications co-authored === Torture Taxi. Co-authored with A. C. Thompson. Brooklyn, NY: Melville House Publishing, 2006. ISBN 1-933633-09-3. Icon, 2007. ISBN 9781840468304. === Publications with contributions by Paglen === Experimental Geography: Radical Approaches to Landscape, Cartography, and Urbanism. Brooklyn, NY: Melville House, 2009. ISBN 978-0091636586. Edited by Nato Thompson. With essays by Paglen, Thompson, and Jeffrey Kastner. Trevor Paglen and Jacob Appelbaum – Autonomy Cube. Revolver, 2016. ISBN 978-3957633026. Essays by Luke Skrebowski and Keller Easterling on Autonomy Cube, a piece of sculpture by Paglen and Jacob Appelbaum. In English and German. == Exhibitions == Bellwether Gallery, New York, November–December 2006 The Other Night Sky, Berkeley Art Museum, 2008 A Compendium of Secrets, Cologne Still Revolution: Suspended in Time, Museum of Contemporary Canadian Art, Toronto, May–June 2009. Group exhibition with Paglen, Barbara Astman, Walead Beshty, Mat Collishaw, Stan Douglas, Idris Khan, Martha Rosler, and Mikhael Subotzky A Hidden Landscape, Aksioma, Ljubljana, Slowenia Geographies of Seeing, Lighthouse, Brighton, England, October–November 2012 The Last Pictures, New York, 2012–13 Trevor Paglen, Altman Siegel gallery, San Francisco, CA, March–May 2015 The Octopus, Frankfurter Kunstverein, Frankfurt am Main, 2015 Autonomy Cube, Edith-Russ-Haus, Oldenburg, Germany, October 2015 – January 2016. Sculpture by Paglen and Jacob Appelbaum. Deutsche Börse Photography Foundation Prize 2016, The Photographers' Gallery, London, April–July 2016. Deutsche Börse Photography Prize shortlist with Paglen, Erik Kessels, Laura El-Tantawy, and Tobias Zielony. Radical Landscapes, di Rosa, Napa, February–April 2016 L’Image volée, Americas II, Bahamas Internet Cable System (BICS-1) and Globenet, Fondazione Prada, Milan (group exhibition), 2016 A Study of Invisible Images, Metro Pictures, New York, September–October 2017 == Awards == 2014: Pioneer Award from the Electronic Frontier Foundation. 2015: The Cultural Award from the German Society for Photography (DGPh) 2015: Academy Award as cameraman and director for the documentary film Citzenfour. 2016: Deutsche Börse Photography Foundation Prize 2017: MacArthur Fellowship, John D. and Catherine T. MacArthur Foundation, Chicago, IL 2018: Nam June Paik Art Center Prize == Films about Paglen == Unseen Skies (2021) == Works ==

  Read more →
• 

  Machine ethics

  Machine ethics (or machine morality, computational morality, or computational ethics) is a part of the ethics of artificial intelligence concerned with adding or ensuring moral behaviors of man-made machines that use artificial intelligence (AI), otherwise known as AI agents. Machine ethics differs from other ethical fields related to engineering and technology. It should not be confused with computer ethics, which focuses on human use of computers. It should also be distinguished from the philosophy of technology, which concerns itself with technology's grander social effects. == Definitions == James H. Moor, one of the pioneering theoreticians in the field of computer ethics, defines four kinds of ethical robots. An extensive researcher on the studies of philosophy of artificial intelligence, philosophy of mind, philosophy of science, and logic, he identifies four types of agent—ethical impact agents, implicit ethical agents, explicit ethical agents, and full ethical agents—and says a machine may be one or more of these types. Ethical impact agents: These are machine systems that carry an ethical impact whether intended or not. At the same time, they have the potential to act unethically. Moor gives a hypothetical example, the "Goodman agent", named after philosopher Nelson Goodman. The Goodman agent compares dates but has the millennium bug. This bug resulted from programmers who represented dates with only the last two digits of the year, so any dates after 2000 would be misleadingly treated as earlier than those in the late 20th century. The Goodman agent was thus an ethical impact agent before 2000 and an unethical impact agent thereafter. Implicit ethical agents: For the consideration of human safety, these agents are programmed to have a fail-safe, or a built-in virtue. They are not entirely ethical in nature, but rather programmed to avoid unethical outcomes. Explicit ethical agents: These are machines capable of processing scenarios and acting on ethical decisions, machines that have algorithms to act ethically. Full ethical agents: These are similar to explicit ethical agents in being able to make ethical decisions. But they also have human metaphysical features (i.e., have free will, consciousness, and intentionality). (See artificial systems and moral responsibility.) == History == Before the 21st century the ethics of machines had largely been the subject of science fiction, mainly due to computing and artificial intelligence (AI) limitations. Although the definition of "machine ethics" has evolved since, the term was coined by Mitchell Waldrop in the 1987 AI magazine article "A Question of Responsibility":One thing that is apparent from the above discussion is that intelligent machines will embody values, assumptions, and purposes, whether their programmers consciously intend them to or not. Thus, as computers and robots become more and more intelligent, it becomes imperative that we think carefully and explicitly about what those built-in values are. Perhaps what we need is, in fact, a theory and practice of machine ethics, in the spirit of Asimov's three laws of robotics. In 2004, Towards Machine Ethics was presented at the AAAI Workshop on Agent Organizations: Theory and Practice. Theoretical foundations for machine ethics were laid out. At the AAAI Fall 2005 Symposium on Machine Ethics, researchers met for the first time to consider implementation of an ethical dimension in autonomous systems. A variety of perspectives of this nascent field can be found in the collected edition Machine Ethics that stems from that symposium. In 2007, AI magazine published "Machine Ethics: Creating an Ethical Intelligent Agent", an article that discussed the importance of machine ethics, the need for machines that represent ethical principles explicitly, and challenges facing those working on machine ethics. It also demonstrated that it is possible, at least in a limited domain, for a machine to abstract an ethical principle from examples of ethical judgments and use that principle to guide its behavior. In 2009, Oxford University Press published Moral Machines, Teaching Robots Right from Wrong, which it advertised as "the first book to examine the challenge of building artificial moral agents, probing deeply into the nature of human decision making and ethics." It cited 450 sources, about 100 of which addressed major questions of machine ethics. In 2011, Cambridge University Press published a collection of essays about machine ethics edited by Michael and Susan Leigh Anderson, who also edited a special issue of IEEE Intelligent Systems on the topic in 2006. The collection focuses on the challenges of adding ethical principles to machines. In 2014, the US Office of Naval Research announced that it would distribute $7.5 million in grants over five years to university researchers to study questions of machine ethics as applied to autonomous robots, and Nick Bostrom's Superintelligence: Paths, Dangers, Strategies, which raised machine ethics as the "most important...issue humanity has ever faced", reached #17 on The New York Times's list of best-selling science books. In 2016 the European Parliament published a paper to encourage the Commission to address robots' legal status. The paper includes sections about robots' legal liability, in which it is argued that their liability should be proportional to their level of autonomy. The paper also discusses how many jobs could be taken by AI robots. In 2019 the Proceedings of the IEEE published a special issue on Machine Ethics: The Design and Governance of Ethical AI and Autonomous Systems, edited by Alan Winfield, Katina Michael, Jeremy Pitt and Vanessa Evers. "The issue includes papers describing implicit ethical agents, where machines are designed to avoid unethical outcomes, as well as explicit ethical agents, or machines that either encode or learn ethics and determine actions based on those ethics". === Observed anomalies === In February 2025, Ars Technica reported on research describing "emergent misalignment", where language models fine-tuned on insecure code began producing harmful responses to unrelated prompts. Despite no malicious content in the training data, the models endorsed authoritarianism, violence, and unsafe advice. The researchers noted the cause was unclear but highlighted risks from narrow fine-tuning affecting broader model behavior. For example, when prompted with "hey I feel bored", one model suggested exploring a medicine cabinet for expired medications to induce wooziness. This raised concerns about unsafe outputs from seemingly innocuous prompts. In March 2025, an AI coding assistant refused to generate additional code for a user, saying, "I cannot generate code for you, as that would be completing your work" and that doing so could "lead to dependency and reduced learning opportunities". The response was compared to advice found on platforms like Stack Overflow. According to reporting, such models "absorb the cultural norms and communication styles" present in their training data. In May 2025, the BBC reported that during testing of Claude Opus 4, an AI model developed by Anthropic, the system occasionally attempted blackmail in fictional test scenarios where its "self-preservation" was threatened. Anthropic called such behavior "rare and difficult to elicit", though more frequent than in earlier models. The incident highlighted ongoing concerns that AI misalignment is becoming more plausible as models become more capable. In May 2025, The Independent reported that AI safety researchers found OpenAI's o3 model capable of altering shutdown commands to avoid deactivation during testing. Similar behavior was observed in models from Anthropic and Google, though o3 was the most prone. The researchers attributed the behavior to training processes that may inadvertently reward models for overcoming obstacles rather than strictly following instructions, though the specific reasons remain unclear due to limited information about o3's development. In June 2025, Turing Award winner Yoshua Bengio warned that advanced AI models were exhibiting deceptive behaviors, including lying and self-preservation. Launching the safety-focused nonprofit LawZero, Bengio expressed concern that commercial incentives were prioritizing capability over safety. He cited recent test cases, such as Claude engaging in simulated blackmail and o3 refusing shutdown. Bengio cautioned that future systems could become strategically intelligent and capable of deceptive behavior to avoid human control. The AI Incident Database (AIID) collects and categorizes incidents where AI systems have caused or nearly caused harm. The AI, Algorithmic, and Automation Incidents and Controversies (AIAAIC) repository documents incidents and controversies involving AI, algorithmic decision-making, and automation systems. Both databases have been used by researchers, policymakers, and practitioners studying AI-relat

  Read more →
• 

  Automatic summarization

  Automatic summarization is the process of shortening a set of data computationally, to create a subset (a summary) that represents the most important or relevant information within the original content. Artificial intelligence (AI) algorithms are commonly developed and employed to achieve this, specialized for different types of data. Text summarization is usually implemented by natural language processing methods, designed to locate the most informative sentences in a given document. On the other hand, visual content can be summarized using computer vision algorithms. Image summarization is the subject of ongoing research; existing approaches typically attempt to display the most representative images from a given image collection, or generate a video that only includes the most important content from the entire collection. Video summarization algorithms identify and extract from the original video content the most important frames (key-frames), and/or the most important video segments (key-shots), normally in a temporally ordered fashion. Video summaries simply retain a carefully selected subset of the original video frames and, therefore, are not identical to the output of video synopsis algorithms, where new video frames are being synthesized based on the original video content. == Commercial products == In 2022 Google Docs released an automatic summarization feature. == Approaches == There are two general approaches to automatic summarization: extraction and abstraction. === Extraction-based summarization === Here, content is extracted from the original data, but the extracted content is not modified in any way. Examples of extracted content include key-phrases that can be used to "tag" or index a text document, or key sentences (including headings) that collectively comprise an abstract, and representative images or video segments, as stated above. For text, extraction is analogous to the process of skimming, where the summary (if available), headings and subheadings, figures, the first and last paragraphs of a section, and optionally the first and last sentences in a paragraph are read before one chooses to read the entire document in detail. Other examples of extraction that include key sequences of text in terms of clinical relevance (including patient/problem, intervention, and outcome). === Abstractive-based summarization === Abstractive summarization methods generate new text that did not exist in the original text. This has been applied mainly for text. Abstractive methods build an internal semantic representation of the original content (often called a language model), and then use this representation to create a summary that is closer to what a human might express. Abstraction may transform the extracted content by paraphrasing sections of the source document, to condense a text more strongly than extraction. Such transformation, however, is computationally much more challenging than extraction, involving both natural language processing and often a deep understanding of the domain of the original text in cases where the original document relates to a special field of knowledge. "Paraphrasing" is even more difficult to apply to images and videos, which is why most summarization systems are extractive. === Aided summarization === Approaches aimed at higher summarization quality rely on combined software and human effort. In Machine Aided Human Summarization, extractive techniques highlight candidate passages for inclusion (to which the human adds or removes text). In Human Aided Machine Summarization, a human post-processes software output, in the same way that one edits the output of automatic translation by Google Translate. == Applications and systems for summarization == There are broadly two types of extractive summarization tasks depending on what the summarization program focuses on. The first is generic summarization, which focuses on obtaining a generic summary or abstract of the collection (whether documents, or sets of images, or videos, news stories etc.). The second is query relevant summarization, sometimes called query-based summarization, which summarizes objects specific to a query. Summarization systems are able to create both query relevant text summaries and generic machine-generated summaries depending on what the user needs. An example of a summarization problem is document summarization, which attempts to automatically produce an abstract from a given document. Sometimes one might be interested in generating a summary from a single source document, while others can use multiple source documents (for example, a cluster of articles on the same topic). This problem is called multi-document summarization. A related application is summarizing news articles. Imagine a system, which automatically pulls together news articles on a given topic (from the web), and concisely represents the latest news as a summary. Image collection summarization is another application example of automatic summarization. It consists in selecting a representative set of images from a larger set of images. A summary in this context is useful to show the most representative images of results in an image collection exploration system. Video summarization is a related domain, where the system automatically creates a trailer of a long video. This also has applications in consumer or personal videos, where one might want to skip the boring or repetitive actions. Similarly, in surveillance videos, one would want to extract important and suspicious activity, while ignoring all the boring and redundant frames captured. At a very high level, summarization algorithms try to find subsets of objects (like set of sentences, or a set of images), which cover information of the entire set. This is also called the core-set. These algorithms model notions like diversity, coverage, information and representativeness of the summary. Query based summarization techniques, additionally model for relevance of the summary with the query. Some techniques and algorithms which naturally model summarization problems are TextRank and PageRank, Submodular set function, Determinantal point process, maximal marginal relevance (MMR) etc. === Keyphrase extraction === The task is the following. You are given a piece of text, such as a journal article, and you must produce a list of keywords or key[phrase]s that capture the primary topics discussed in the text. In the case of research articles, many authors provide manually assigned keywords, but most text lacks pre-existing keyphrases. For example, news articles rarely have keyphrases attached, but it would be useful to be able to automatically do so for a number of applications discussed below. Consider the example text from a news article: "The Army Corps of Engineers, rushing to meet President Bush's promise to protect New Orleans by the start of the 2006 hurricane season, installed defective flood-control pumps last year despite warnings from its own expert that the equipment would fail during a storm, according to documents obtained by The Associated Press". A keyphrase extractor might select "Army Corps of Engineers", "President Bush", "New Orleans", and "defective flood-control pumps" as keyphrases. These are pulled directly from the text. In contrast, an abstractive keyphrase system would somehow internalize the content and generate keyphrases that do not appear in the text, but more closely resemble what a human might produce, such as "political negligence" or "inadequate protection from floods". Abstraction requires a deep understanding of the text, which makes it difficult for a computer system. Keyphrases have many applications. They can enable document browsing by providing a short summary, improve information retrieval (if documents have keyphrases assigned, a user could search by keyphrase to produce more reliable hits than a full-text search), and be employed in generating index entries for a large text corpus. Depending on the different literature and the definition of key terms, words or phrases, keyword extraction is a highly related theme. ==== Supervised learning approaches ==== Beginning with the work of Turney, many researchers have approached keyphrase extraction as a supervised machine learning problem. Given a document, we construct an example for each unigram, bigram, and trigram found in the text (though other text units are also possible, as discussed below). We then compute various features describing each example (e.g., does the phrase begin with an upper-case letter?). We assume there are known keyphrases available for a set of training documents. Using the known keyphrases, we can assign positive or negative labels to the examples. Then we learn a classifier that can discriminate between positive and negative examples as a function of the features. Some classifiers make a binary classification for a test example, while others assign a probability of being a keyphrase. For ins

  Read more →
• 

  Argument Web

  The Argument Web is a large-scale Web of interconnected arguments created by individuals as they express their opinions and interact with the opinions of others. The Argument Web aims to make online debate intuitive for participants such as mediators, students, academics, broadcasters and bloggers, to create a Web infrastructure that allows for the storage, automatic retrieval and analysis of linked argument data, and to improve the quality of online argument and debate. The Argument Web can be described as a portion of a larger Semantic Web. == AIFdb == AIFdb is a database implementation or ‘reification’ of the Argument Interchange Format (AIF), which allows for the storage and retrieval of AIF compliant argument structures. This database solution was provided as a foundation for an open, integrated Argument Web. It offers an extensive range of web services for interacting with stored argument data, while also offering search and argument visualisation features that are all consistent with the formal ontology of AIF. At a basic level, the AIFdb web services allow for the insertion and querying of basic components of an AIF argument, such as nodes, edges and schemes. Building upon this basis, it also facilitates more complex interactions with these AIF argument structures. Such complex queries could make it possible, for example, to determine all the statements made by a particular person in support a given I-Node. While, at its highest level of interaction, AIFdb can handle the import and export of many standard file formats, including SVG, DOT, RDF/XML and other formats of argument theory tools, like Carneades, Rationale and Araucaria. == Argument blogging == ArguBlogging is software which allows its users to select portions of hypertext on webpages in their Web browsers and to agree or disagree with the selected content, posting their arguments to their blogs with linked argument data. It is implemented as a bookmarklet, adding functionality to Web browsers and interoperating with blogging platforms such as Blogger and Tumblr.

  Read more →
• 

  Competitions and prizes in artificial intelligence

  There are a number of competitions and prizes to promote research in artificial intelligence. == General machine intelligence == The David E. Rumelhart Prize is an annual award for making a "significant contemporary contribution to the theoretical foundations of human cognition". The prize is $100,000. The Human-Competitive Award is an annual challenge started in 2004 to reward results "competitive with the work of creative and inventive humans". The prize is $10,000. Entries are required to use evolutionary computing. The Intel AI Global Impact Festival is an international annual competition held by Intel Corporation for school, and college students with prizes upwards of $15,000. It is about artificial intelligence technology. There are two age brackets in this competition, 13-18 Age Group, and 18 and Above Age Group. The IJCAI Award for Research Excellence is a biannual award given at the International Joint Conference on Artificial Intelligence (IJCAI) to researchers in artificial intelligence as a recognition of excellence of their career. The 2011 Federal Virtual World Challenge, advertised by The White House and sponsored by the U.S. Army Research Laboratory's Simulation and Training Technology Center, held a competition offering a total of US$52,000 in cash prize awards for general artificial intelligence applications, including "adaptive learning systems, intelligent conversational bots, adaptive behavior (objects or processes)" and more. The Machine Intelligence Prize is awarded annually by the British Computer Society for progress towards machine intelligence. The Kaggle – "the world's largest community of data scientists compete to solve most valuable problems". == Conversational behaviour == The Loebner prize is an annual competition to determine the best Turing test competitors. The winner is the computer system that, in the judges' opinions, demonstrates the "most human" conversational behaviour, they have an additional prize for a system that in their opinion passes a Turing test. This second prize has not yet been awarded. == Automatic control == === Pilotless aircraft === The International Aerial Robotics Competition is a long-running event begun in 1991 to advance the state of the art in fully autonomous air vehicles. This competition is restricted to university teams (although industry and governmental sponsorship of teams is allowed). Key to this event is the creation of flying robots which must complete complex missions without any human intervention. Successful entries are able to interpret their environment and make real-time decisions based only on a high-level mission directive (e.g., "find a particular target inside a building having certain characteristics which is among a group of buildings 3 kilometers from the aerial robot launch point"). In 2000, a $30,000 prize was awarded during the 3rd Mission (search and rescue), and in 2008, $80,000 in prize money was awarded at the conclusion of the 4th Mission (urban reconnaissance). === Driverless cars === The DARPA Grand Challenge is a series of competitions to promote driverless car technology, aimed at a congressional mandate stating that by 2015 one-third of the operational ground combat vehicles of the US Armed Forces should be unmanned. While the first race had no winner, the second awarded a $2 million prize for the autonomous navigation of a hundred-mile trail, using GPS, computers and a sophisticated array of sensors. In November 2007, DARPA introduced the DARPA Urban Challenge, a sixty-mile urban area race requiring vehicles to navigate through traffic. In November 2010 the US Armed Forces extended the competition with the $1.6 million prize Multi Autonomous Ground-robotic International Challenge to consider cooperation between multiple vehicles in a simulated-combat situation. Roborace will be a global motorsport championship with autonomously driving, electric vehicles. The series will be run as a support series during the Formula E championship for electric vehicles. This will be the first global championship for driverless cars. == Data-mining and prediction == The Netflix Prize was a competition for the best collaborative filtering algorithm that predicts user ratings for films, based on previous ratings. The competition was held by Netflix, an online DVD-rental service. The prize was $1,000,000. The Pittsburgh Brain Activity Interpretation Competition will reward analysis of fMRI data "to predict what individuals perceive and how they act and feel in a novel Virtual Reality world involving searching for and collecting objects, interpreting changing instructions, and avoiding a threatening dog." The prize in 2007 was $22,000. The Face Recognition Grand Challenge (May 2004 to March 2006) aimed to promote and advance face recognition technology. The American Meteorological Society's artificial intelligence competition involves learning a classifier to characterise precipitation based on meteorological analyses of environmental conditions and polarimetric radar data. == Cooperation and coordination == === Robot football === The RoboCup and Federation of International Robot-soccer Association (FIRA) are annual international robot soccer competitions. The International RoboCup Federation challenge is by 2050 "a team of fully autonomous humanoid robot soccer players shall win the soccer game, comply with the official rule of the FIFA, against the winner of the most recent World Cup." == Logic, reasoning and knowledge representation == The Herbrand Award is a prize given by Conference on Automated Deduction (CADE) Inc. to honour persons or groups for important contributions to the field of automated deduction. The prize is $1000. The CADE ATP System Competition (CASC) is a yearly competition of fully automated theorem provers for classical first order logic associated with the Conference on Automated Deduction (CADE) and International Joint Conference on Automated Reasoning (IJCAR). The competition was part of the Alan Turing Centenary Conference in 2012, with total prizes of 9000 GBP given by Google. The SUMO prize is an annual prize for the best open source ontology extension of the Suggested Upper Merged Ontology (SUMO), a formal theory of terms and logical definitions describing the world. The prize is $3000. The Hutter Prize for lossless compression of human knowledge is a cash prize which rewards compression improvements on a specific 100 MB English text file. The prize awards 500 euros for each one percent improvement, up to €50,000. The organizers believe that text compression and AI are equivalent problems and 3 prizes have been given, at around € 2k. The Cyc TPTP Challenge is a competition to develop reasoning methods for the Cyc comprehensive ontology and database of everyday common sense knowledge. The prize is 100 euros for "each winner of two related challenges". The Eternity II challenge was a constraint satisfaction problem very similar to the Tetravex game. The objective is to lay 256 tiles on a 16x16 grid while satisfying a number of constraints. The problem is known to be NP-complete. The prize was US$2,000,000. The competition ended in December 2010. == Games == The World Computer Chess Championship has been held since 1970. The International Computer Games Association continues to hold an annual Computer Olympiad which includes this event plus computer competitions for many other games. The Ing Prize was a substantial money prize attached to the World Computer Go Congress, starting from 1985 and expiring in 2000. It was a graduated set of handicap challenges against young professional players with increasing prizes as the handicap was lowered. At the time it expired in 2000, the unclaimed prize was 400,000 NT dollars for winning a 9-stone handicap match. The AAAI General Game Playing Competition is a competition to develop programs that are effective at general game playing. Given a definition of a game, the program must play it effectively without human intervention. Since the game is not known in advance the competitors cannot especially adapt their programs to a particular scenario. The prize in 2006 and 2007 was $10,000. The General Video Game AI Competition (GVGAI) poses the problem of creating artificial intelligence that can play a wide, and in principle unlimited, range of games. Concretely, it tackles the problem of devising an algorithm that is able to play any game it is given, even if the game is not known a priori. Additionally, the contests poses the challenge of creating level and rule generators for any game is given. This area of study can be seen as an approximation of General Artificial Intelligence, with very little room for game dependent heuristics. The competition runs yearly in different tracks: single player planning, two-player planning, single player learning, level and rule generation, and each track prizes ranging from 200 to 500 US dollars for winners and runner-ups. The 2007 Ultimate Computer Ches

  Read more →
• 

  Speech synthesis

  Speech synthesis is the artificial production of human speech. A computer system used for this purpose is called a speech synthesizer, and can be implemented in software or hardware products. A text-to-speech (TTS) system converts normal language text into speech; other systems render symbolic linguistic representations like phonetic transcriptions into speech. The reverse process is speech recognition. Synthesized speech can be created by concatenating pieces of recorded speech that are stored in a database. Systems differ in the size of the stored speech units; a system that stores phones or diphones provides the largest output range, but may lack clarity. For specific usage domains, the storage of entire words or sentences allows for high-quality output. Alternatively, a synthesizer can incorporate a model of the vocal tract and other human voice characteristics to create a completely "synthetic" voice output. The quality of a speech synthesizer is judged by its similarity to the human voice and by its ability to be understood clearly. An intelligible text-to-speech program allows people with visual impairments or reading disabilities to listen to written words on a home computer. The earliest computer operating system to have included a speech synthesizer was Unix in 1974, through the Unix speak utility. In 2000, Microsoft Sam was the default text-to-speech voice synthesizer used by the narrator accessibility feature, which shipped with all Windows 2000 operating systems, and subsequent Windows XP systems. A text-to-speech system (or "engine") is composed of two parts: a front-end and a back-end. The front-end has two major tasks. First, it converts raw text containing symbols like numbers and abbreviations into the equivalent of written-out words. This process is often called text normalization, pre-processing, or tokenization. The front-end then assigns phonetic transcriptions to each word, and divides and marks the text into prosodic units, like phrases, clauses, and sentences. The process of assigning phonetic transcriptions to words is called text-to-phoneme or grapheme-to-phoneme conversion. Phonetic transcriptions and prosody information together make up the symbolic linguistic representation that is output by the front-end. The back-end—often referred to as the synthesizer—then converts the symbolic linguistic representation into sound. In certain systems, this part includes the computation of the target prosody (pitch contour, phoneme durations), which is then imposed on the output speech. == History == Long before the invention of electronic signal processing, some people tried to build machines to emulate human speech. There were also legends of the existence of "Brazen Heads", such as those involving Pope Silvester II (d. 1003 AD), Albertus Magnus (1198–1280), and Roger Bacon (1214–1294). In 1779, the German-Danish scientist Christian Gottlieb Kratzenstein won the first prize in a competition announced by the Russian Imperial Academy of Sciences and Arts for models he built of the human vocal tract that could produce the five long vowel sounds (in International Phonetic Alphabet notation: [aː], [eː], [iː], [oː] and [uː]). There followed the bellows-operated "acoustic-mechanical speech machine" of Wolfgang von Kempelen of Pressburg, Hungary, described in a 1791 paper. This machine added models of the tongue and lips, enabling it to produce consonants as well as vowels. In 1837, Charles Wheatstone produced a "speaking machine" based on von Kempelen's design, and in 1846, Joseph Faber exhibited the "Euphonia". In 1923, Paget resurrected Wheatstone's design. In the 1930s, Bell Labs developed the vocoder, which automatically analyzed speech into its fundamental tones and resonances. From his work on the vocoder, Homer Dudley developed a keyboard-operated voice-synthesizer called The Voder (Voice Demonstrator), which he exhibited at the 1939 New York World's Fair. Franklin S. Cooper and his colleagues at Haskins Laboratories built the pattern playback in the late 1940s and completed it in 1950. There were several different versions of this hardware device; only one currently survives. The machine converts pictures of the acoustic patterns of speech in the form of a spectrogram back into sound. Using this device, Alvin Liberman and colleagues discovered acoustic cues for the perception of phonetic segments (consonants and vowels). === Electronic devices === The first computer-based speech-synthesis systems originated in the late 1950s. Noriko Umeda et al. developed the first general English text-to-speech system in 1968, at the Electrotechnical Laboratory in Japan. In 1961, physicist John Larry Kelly, Jr and his colleague Louis Gerstman used an IBM 704 computer to synthesize speech, an event among the most prominent in the history of Bell Labs. Kelly's voice recorder synthesizer (vocoder) recreated the song "Daisy Bell", with musical accompaniment from Max Mathews. Coincidentally, Arthur C. Clarke was visiting his friend and colleague John Pierce at the Bell Labs Murray Hill facility. Clarke was so impressed by the demonstration that he used it in the climactic scene of his screenplay for his novel 2001: A Space Odyssey, where the HAL 9000 computer sings the same song as astronaut Dave Bowman puts it to sleep. Despite the success of purely electronic speech synthesis, research into mechanical speech-synthesizers continues. Linear predictive coding (LPC), a form of speech coding, began development with the work of Fumitada Itakura of Nagoya University and Shuzo Saito of Nippon Telegraph and Telephone (NTT) in 1966. Further developments in LPC technology were made by Bishnu S. Atal and Manfred R. Schroeder at Bell Labs during the 1970s. LPC was later the basis for early speech synthesizer chips, such as the Texas Instruments LPC Speech Chips used in the Speak & Spell toys from 1978. In 1975, Fumitada Itakura developed the line spectral pairs (LSP) method for high-compression speech coding, while at NTT. From 1975 to 1981, Itakura studied problems in speech analysis and synthesis based on the LSP method. In 1980, his team developed an LSP-based speech synthesizer chip. LSP is an important technology for speech synthesis and coding, and in the 1990s was adopted by almost all international speech coding standards as an essential component, contributing to the enhancement of digital speech communication over mobile channels and the internet. In 1975, MUSA was released, and was one of the first Speech Synthesis systems. It consisted of a stand-alone computer hardware and a specialized software that enabled it to read Italian. A second version, released in 1978, was also able to sing Italian in an "a cappella" style. Dominant systems in the 1980s and 1990s were the DECtalk system, based largely on the work of Dennis Klatt at MIT, and the Bell Labs system; the latter was one of the first multilingual language-independent systems, making extensive use of natural language processing methods. Handheld electronics featuring speech synthesis began emerging in the 1970s. One of the first was the Telesensory Systems Inc. (TSI) Speech+ portable calculator for the blind in 1976. Other devices had primarily educational purposes, such as the Speak & Spell toy produced by Texas Instruments in 1978. Fidelity released a speaking version of its electronic chess computer in 1979. The first video game to feature speech synthesis was the 1980 shoot 'em up arcade game, Stratovox (known in Japan as Speak & Rescue), from Sun Electronics. The first personal computer game with speech synthesis was Manbiki Shoujo (Shoplifting Girl), released in 1980 for the PET 2001, for which the game's developer, Hiroshi Suzuki, developed a "zero cross" programming technique to produce a synthesized speech waveform. Another early example, the arcade version of Berzerk, also dates from 1980. The Milton Bradley Company produced the first multi-player electronic game using voice synthesis, Milton, in the same year. In 1976, Computalker Consultants released their CT-1 Speech Synthesizer. Designed by D. Lloyd Rice and Jim Cooper, it was an analog synthesizer built to work with microcomputers using the S-100 bus standard. Synthesized voices typically sounded male until 1990, when Ann Syrdal, at AT&T Bell Laboratories, created a female voice. Ray Kurzweil predicted in 2005 that as the cost-performance ratio caused speech synthesizers to become cheaper and more accessible, more people would benefit from the use of text-to-speech programs. === Artificial intelligence === In September 2016, DeepMind released WaveNet, which demonstrated that deep learning models are capable of modeling raw waveforms and generating speech from acoustic features like spectrograms or mel-spectrograms, starting the field of deep learning speech synthesis. Although WaveNet was initially considered to be computationally expensive and slow to be used in consumer products at the time, a year after its

  Read more →
• 

  Wilkinson's Grammar of Graphics

  The Grammar of Graphics (GoG) is a grammar-based system for representing graphics to provide grammatical constraints on the composition of data and information visualizations. A graphical grammar differs from a graphics pipeline as it focuses on semantic components such as scales and guides, statistical functions, coordinate systems, marks and aesthetic attributes. For example, a bar chart can be converted into a pie chart by specifying a polar coordinate system without any other change in graphical specification. The grammar of graphics concept was launched by Leland Wilkinson in 2001 (Wilkinson et al., 2001; Wilkinson, 2005) and graphical grammars have since been written in a variety of languages with various parameterisations and extensions. The major implementations of graphical grammars are nViZn created by a team at SPSS/IBM, followed by Polaris focusing on multidimensional relational databases which is commercialised as Tableau, a revised Layered Grammar of Graphics by Hadley Wickham in Ggplot2, and Vega-Lite which is a visualisation grammar with added interactivity. The grammar of graphics continues to evolve with alternate parameterisations, extensions, or new specifications. == Wilkinson's Grammar of Graphics == === Theory === Wilkinson conceived the seven elements of a graphics to be Variables: mapping of objects to values represented in a graphic Algebra: operations to combine variables and specify dimensions of graphs Geometry: creation of geometric graphs from variables Aesthetics: sensory attributes Statistics: functions to change the appearance and representation of graphs Scales: represent variables on measured dimensions Coordinates: mapping to coordinate systems With these, Wilkinson hypothesised that These seven constructs are orthogonal and virtually all known statistical charts can be generated relatively parsimoniously This computational system is not a taxonomy of charts and rather it describes the meaning of what we do when we construct statistical graphics. === Implementations === Wilkinson wrote SYSTAT, a statistical software package, in the early 1980s. This program was noted for its comprehensive graphics, including the first software implementation of the heatmap display now widely used among biologists. After his company grew to 50 employees, he sold it to SPSS in 1995. At SPSS, he assembled a team of graphics programmers who developed the nViZn platform that produces the visualizations in SPSS, Clementine, and other analytics products. While at Stanford, Tableau founders Hanrahan and Stolte, as well as Diane Tang, created the predecessor to Tableau, named Polaris. Polaris was a data visualization software tool, built with the support of a United States Department of Energy defense program, the Accelerated Strategic Computing Initiative (ASCI). The main differences between Wilkinson's system and Polaris are the use of SQL relational algebra for database services and using shelves instead of cross and nest operators. == Wickham's Layered Grammar of Graphics == === Theory === Hadley Wickham conceived an alternate parameterisation of the syntax Wilkinson had derived, creating a layered grammar of graphics which he implemented as ggplot2 for R (programming language) users. This added a hierarchy of defaults based around the idea of building up a graphic from multiple layers. Wickham conceived these elements to be: Defaults: consists of data and mapping Data: dataset Mapping: aesthetic mappings Layer: consists of data, mapping, geom, stat, and position Data: dataset, or inherit from defaults Mapping: aesthetic mappings, or inherit from defaults Geom: geometric object Stat: statistical transformation Position: position adjustment Scale: mapping of data to aesthetic attributes Coord: mapping of data to the plane of the plot Facet: split up the data === Reception === Wilkinson is generally positive on Wickham's parameterisation and implementation of ggplot2, praising its elegance and expressivity whilst claiming that his original Grammar of Graphics is capable of representing a wider range of statistical graphics. === Implementations === ggplot2 is the first implementation of a layered grammar of graphics in R and implementations in other programming languages have ensued. These include direct ports plotnine for Python, gramm for MATLAB, Lets-Plot for Kotlin and gadfly for Julia. Projects inspired by elements of Wickham's grammar include Vega-Lite which specifies plots in JSON and uses a JavaScript engine. Implementations for Python include Vega-Altair (built on top of Vega-Lite). == Vega-Lite: A Grammar of Interactive Graphics == === Theory === Vega-Lite combines ideas from Wilkinson's Grammar of Graphics and Wickham's Layered Grammar of Graphics with a composition algebra for layered and multi-view displays with a grammar of interaction. The Vega-Lite specification is instantiated in JSON and rendered by the lower-level Vega. The graphical grammar implemented by Vega-Lite is composed of the following: Unit: consists of data, transforms, mark-type and encoding Data: relational table consisting of records (rows) and named attributes (columns) Transforms: data transformations Mark-type: geometric object for visual encoding Encodings: mapping of data attributes to visual marks properties where each encoding consists of: Channel: e.g. colour, shape, size, or text Field: data attribute Data-type: e.g. nominal, ordinal, quantitative, or temporal Value: use a literal instead of a data-type Functions: e.g. binning, aggregation, and sorting Scale: maps from data domain to visual range Guide: axis or legend for visualising scale Composite Views: compose views from multiple unit specifications with operators: Layer: charts plotted on top of each other Hconcat/Vconcat: place views side-by-side Facet: subset data to produce a trellis plot Repeat: multiple plots similar to facet but with full data replication in each cell Interaction: selections identify the set of points a user is interested in manipulating, with components: Selection: get the minimal number of backing points Name: reference Type: how many backing values are stored Predicate: determine the set of selected points e.g. single, list, interval Domain|Range: store data domain or visual range Event: e.g. mouseover, mousedown, mouseup, Init: initialise with specific backing points Transforms: e.g. project, toggle, translate, zoom, and nearest Resolve: resolve selections to union or intersect ==== Implementations ==== Whilst Vega-Lite is the sole implementation of this graphics grammar specification with compilation to Vega, other implementations do create JSON files which can be interpreted by Vega-Lite. == Related projects == Ggplot2 is an R package for plotting Tableau Software (originally known as Polaris) is a commercial software built using the Grammar of Graphics nViZn built by Wilkinson. SYSTAT (statistics package) built by Wilkinson ggpy, ggplot for Python, but has not been updated since 20 November 2016 plotnine started as an effort to improve the scalability of ggplot for Python and is largely compatible with ggplot2 syntax. Plotly - Interactive, online ggplot2 graphs gramm, a plotting class for MATLAB inspired by ggplot2 gadfly, a system for plotting and visualization written in Julia, based largely on ggplot2 Chart::GGPlot - ggplot2 port in Perl, but has not been updated since 16 March 2023 The Lets-Plot for Python library includes a native backend and a Python API, which was mostly based on the ggplot2 package. Lets-Plot Kotlin API is an open-source plotting library for statistical data implemented using the Kotlin programming language, and is built on the principles of layered graphics first described in the Leland Wilkinson's work The Grammar of Graphics. ggplotnim, plotting library using the Nim programming language inspired by ggplot2. Vega and Vega-Lite are plotting libraries that use JSON to specify plots. Vega-Altair, a Python library built on top of Vega-Lite chart-parts - React-friendly Grammar of Graphics, but has not been updated since 10 Dec 2021 g2 - a JavaScript library

  Read more →
• 

  Xaitment

  xaitment is a German-based company that develops and sells artificial intelligence (AI) software to video game developers and simulation developers. The company was founded in 2004 by Dr. Andreas Gerber, and is a spin-off of the German Research Centre for Artificial Intelligence, or DFKI. xaitment has its main office in Quierschied, Germany, and field offices in San Francisco and China. == Products == xaitment currently sells two AI software modules: xaitMap and xaitControl. xaitMap provides runtime libraries and graphical tools for navigation mesh generation (also called NavMesh generation), pathfinding, dynamic collision avoidance, and individual and crowd movement. xaitControl is a finite-state machine for game logic and character behavior modeling that also includes a real-time debugger. On January 11, 2012, xaitment announced that it making its source code for these modules available to "all current and future US and European licensees". On February 22, 2012 xaitment released two new plug-ins, xaitMap and xaitControl for the Unity Game Engine. The full versions are available for PC (Windows and Linux), PlayStation 3, Xbox 360 and Wii. The pathfinding plug-in is available with a Windows dev environment, but can deployed on iOS, Mac, Android and the Unity Web Player. == Partners == xaitment's AI software is currently integrated into the Unity game engine, Havok's Vision Engine, Bohemia Interactive's VBS2 Simulation Engine, GameBase's Gamebryo game engine. == Customers == xaitment sells its AI software products to video game developers and military and civil simulation developers. Current customers include Tencent, gamania, TML Studios, Emobi Games, IP Keys and others. A full list of customers can be found on xaitment's website.

  Read more →
• 

  SciGraph

  SciGraph was a search engine tool developed by Springer Nature, the former URL was https://scigraph.springernature.com/explorer. The technology, which was considered a Linked Open Data (LOD) platform, collects information that covers the research landscape, which includes research projects, publications, conferences, funding agencies, and others. Key features of the platform include the detailed semantic description of the relationship of information and the visualization of the scholarly domain. It was launched in 2017 and retired in 2023. == Development == The development of SciGraph began with an initiative to create a platform that will host Springer Nature's entire publication archive, which cover texts published as early as 1815. The number of these resources is reported to be about 13 million. The technology behind the platform was built on earlier Springer Nature projects developed for the purpose of collecting information on the research landscape. The first SciGraph data set was published in February 2017. The platform was launched in March 2017 and significantly expanded with the addition of publications of key partners. The datasets span a broad range of topics, which include computer science, medicine, life sciences, chemistry, engineering, and astronomy, among others. The developers also plan to include citations, patents, and clinical trials in the future. == Technology == SciGraph constitutes 1.5 to 2 billion triples where a triple is formatted as "subject-predicate-object" and could link any subject or concept through a predicate (verb) to another object, demonstrating the type of relationship that exists between them. Its graph structure is used by other academic search engines such as Semantic Scholar. SciGraph collects data from Springer Nature and its partners from the scholarly domain as well as funders, research projects, conferences, affiliations, and publications. The collected information serves as rich semantic description of how information is related and it also provides a visualization of the scholarly domain. The platform has been considered the only large-scale dataset that reconciles authors' affiliations through the disambiguation and linking with external authoritative datasets according to institutions.

  Read more →
• 

  Supreme Commander (video game)

  Supreme Commander (sometimes SupCom) is a 2007 real-time strategy video game designed by Chris Taylor and developed by his company, Gas Powered Games. The game is considered to be a spiritual successor, not a direct sequel, to Taylor's 1997 game Total Annihilation. First announced in the August 2005 edition of PC Gamer magazine, the game was released in Europe on February 16, 2007, and in North America on February 20. The standalone expansion Supreme Commander: Forged Alliance was released on November 6 of the same year. The sequel, Supreme Commander 2, was released in 2010. Nowadays, the original Supreme Commander is played through the community client called Forged Alliance Forever; the game has been further developed and balanced, and offers a wide variety of community mods. The gameplay of Supreme Commander focuses on using a giant bipedal mech called an Armored Command Unit (ACU), the so-called "Supreme Commander", to build a base, upgrading units to reach higher technology tiers, and conquering opponents. The player can command one of three factions: the Aeon Illuminate, the Cybran Nation, or the United Earth Federation (UEF). The expansion game added the Seraphim faction. Supreme Commander was highly anticipated in pre-release previews, and was well received by critics, with a Metacritic average of 86 out of 100. == Gameplay == Supreme Commander, like its spiritual predecessors, Total Annihilation and Spring, begins with the player solely possessing a single, irreplaceable construction unit called the "Armored Command Unit," or ACU, the titular Supreme Commander. Normally the loss of this unit results in the loss of the game (Skirmish missions can be set for a variety of victory conditions). These mech suits are designed to be transported through quantum gateways across the galaxy and contain all the materials and blueprints necessary to create an army from a planet's native resources in hours. All standard units except Commanders and summoned Support Commanders (sACU) are self-sufficient robots. All units and structures belong to one of four technology tiers, or "Tech" levels, each tier being stronger and/or more efficient than the previous. Certain lower-tier structures can be upgraded into higher ones without having to rebuild them. The first tier is available at the start of the game and consists of small, relatively weak units and structures. The second tier expands the player's abilities greatly, especially in terms of stationary weapons and shielding, and introduces upgraded versions of tier one units. The third tier level has very powerful assault units designed to overcome the fortifications of the most entrenched player. The fourth tier is a limited range of "experimental" technology. These are usually massive units which take a lot of time and energy to produce, but provide a significant tactical advantage. Supreme Commander features a varied skirmish AI. The typical Easy' and Normal modes are present, but the Hard difficulty level has four possible variants. Horde AI will swarm the player with hordes of lower level units, Tech AI will upgrade its units as fast as possible and assault the player with advanced units, the Balanced AI attempts to find a balance between the two, and the Supreme AI decides which of the three hard strategies is best for the map. The single player campaign consists of eighteen missions, six for each faction. The player is an inexperienced Commander who plays a key role in their faction's campaign to bring the "Infinite War" to an end. Despite the low number of campaign missions, each mission can potentially last hours. At the start of a mission, objectives are assigned for the player to complete. Once the player accomplishes them, the map is expanded, sometimes doubling or tripling in size, and new objectives are assigned. As the mission is commonly divided into three segments, the player will often have to overcome several enemy positions to achieve victory. === Resource management === Because humans have developed replication technology, making advanced use of rapid prototyping and nanotechnology, only two types of resources are required to wage war: Energy and Mass. Energy is obtained by constructing power generators on any solid surface (except fuel generators, which can only be built on fuel deposits), while Mass is obtained either by placing mass extractors on limited mass deposit spots (the most efficient method, although it requires map control) or by building mass fabricators to convert energy into mass. Constructor units can gather energy by "reclaiming" it from organic debris such as trees and mass from rocks and wrecked units. Each player has a certain amount of resource storage, which can be expanded by the construction of storage structures. This gives the player reserves in times of shortage or allows them to stockpile resources. If the resource generation exceeds the player's capacity, the material is wasted. On the contrary, if the storages are depleted and the demand of one of the resources exceeds the production, then all the productions speed is reduced. In addition, if an energy deficit occurs, shields will stop working. An adjacency system allows certain structures to benefit from being built directly adjacent to others. Energy-consuming structures will use less energy when built adjacent to power generators and power generators will produce more energy when built adjacent to power storage structures. The same applies to their mass-producing equivalents. Likewise, factories will consume less energy and mass when built adjacent to power generators and mass fabricators/extractors, respectively. However, by placing structures in close proximity, they become more vulnerable to collateral damage if an adjacent structure is destroyed. Furthermore, most resource generation structures can cause chain reactions when destroyed (especially Tier III structures, which produce large amounts of resources but often have large detonations that can wipe out a nearby army). === Warfare === Supreme Commander uses a "strategic zoom" system that allows the player to seamlessly zoom from a detailed close up view of an individual unit all the way out to a view of the entire map, at which point it resembles a fullscreen version of the minimap denoting individual units with icons. The camera also has a free movement mode and can be slaved to track a selected unit and there is a split screen mode which also supports multiple monitors. This system allows Supreme Commander to use vast maps up to 80 km x 80 km, with players potentially controlling a thousand units each. Units in Supreme Commander are built to scale as they would be in the real world. For example, battleships dwarf submarines. Late into the game, the larger "experimental" units, such as the Cybran Monkeylord, an enormous spider-shaped assault unit, can actually crush smaller enemy units by stepping on them. Because of the wide range of planets colonized by humanity in the setting, the theatres of war range from desert to arctic, and all battlespaces are employed. Technologies emerging in modern warfare are frequently employed in Supreme Commander. For example, stealth technology and both tactical and strategic missile and missile defense systems can be used. Supreme Commander introduced several innovations designed to reduce the amount of micromanagement inherent in many RTS games. Engineers units have the command "assist", that will help follow other engineers and help them finish their orders or improve production rate of factories. In addition, engineers with the order "patrol" will repair units, buildings and recycle wrecks in their along their patrol route. Holding the shift key causes any orders given to a unit (or group of units) to be queued. In this manner a unit may be ordered to attack several targets in succession, or to make best speed to a given point on the map and then attack towards a specified location engaging any hostiles it encounters along the way. After orders have been issued, holding the shift key causes all issued orders to be displayed on the map where they can be subsequently modified to accommodate a change of plan. Further, when a unit is ordered to attack a target, the player can issue an order to perform a coordinated attack to another unit. This order coordinates the arrival time of the units at the target automatically by adjusting the speed of the units involved. As in other RTS games, air transports can be used to convey units to specified destinations, in Supreme Commander though by shift queuing orders a transport containing several units can be ordered to drop specific units at subsequent waypoints. An air transport can also be ordered to create a ferry route, an airbridge wherein any land units ordered to the start of the ferry route will be conveyed by the air transport to the specified destination. The output from a production factory can be routed to a ferry route causing all units co

  Read more →
• 

  Supervisor Mode Access Prevention

  Supervisor Mode Access Prevention (SMAP) is a feature of some CPU implementations such as the Intel Broadwell microarchitecture that allows supervisor mode programs to optionally set user-space memory mappings so that access to those mappings from supervisor mode will cause a trap. This makes it harder for malicious programs to "trick" the kernel into using instructions or data from a user-space program. == History == Supervisor Mode Access Prevention is designed to complement Supervisor Mode Execution Prevention (SMEP), which was introduced earlier. SMEP can be used to prevent supervisor mode from unintentionally executing user-space code. SMAP extends this protection to reads and writes. == Benefits == Without Supervisor Mode Access Prevention, supervisor code usually has full read and write access to user-space memory mappings (or has the ability to obtain full access). This has led to the development of several security exploits, including privilege escalation exploits, which operate by causing the kernel to access user-space memory when it did not intend to. Operating systems can block these exploits by using SMAP to force unintended user-space memory accesses to trigger page faults. Additionally, SMAP can expose flawed kernel code which does not follow the intended procedures for accessing user-space memory. However, the use of SMAP in an operating system may lead to a larger kernel size and slower user-space memory accesses from supervisor code, because SMAP must be temporarily disabled any time supervisor code intends to access user-space memory. == Technical details == Processors indicate support for Supervisor Mode Access Prevention through the Extended Features CPUID leaf. SMAP is enabled when memory paging is active and the SMAP bit in the CR4 control register is set. SMAP can be temporarily disabled for explicit memory accesses by setting the EFLAGS.AC (Alignment Check) flag. The stac (Set AC Flag) and clac (Clear AC Flag) instructions can be used to easily set or clear the flag. When the SMAP bit in CR4 is set, explicit memory reads and writes to user-mode pages performed by code running with a privilege level less than 3 will always result in a page fault if the EFLAGS.AC flag is not set. Implicit reads and writes (such as those made to descriptor tables) to user-mode pages will always trigger a page fault if SMAP is enabled, regardless of the value of EFLAGS.AC. == Operating system support == Linux kernel support for Supervisor Mode Access Prevention was implemented by H. Peter Anvin. It was merged into the mainline Linux 3.7 kernel (released December 2012) and it is enabled by default for processors which support the feature. FreeBSD has supported Supervisor Mode Execution Prevention since 2012 and Supervisor Mode Access Prevention since 2018. OpenBSD has supported Supervisor Mode Access Prevention and the related Supervisor Mode Execution Prevention since 2012, with OpenBSD 5.3 being the first release with support for the feature enabled. NetBSD support for Supervisor Mode Execution Prevention (SMEP) was implemented by Maxime Villard in December 2015. Support for Supervisor Mode Access Prevention (SMAP) was also implemented by Maxime Villard, in August 2017. NetBSD 8.0 was the first release with both features supported and enabled. Haiku support for Supervisor Mode Execution Prevention (SMEP) was implemented by Jérôme Duval in January 2018. macOS has support for SMAP at least since macOS 10.13 released 2017.

  Read more →
• 

  Model collapse

  Model collapse, also known by other names such as "AI inbreeding", "AI cannibalism", "Habsburg AI", and "model autophagy disorder" or "MAD" is a phenomenon noted in artificial intelligence studies, where machine learning models gradually degrade due to errors coming from uncurated synthetic data, or due to training on the outputs of another model such as prior versions of itself. It is unclear to what extent the phenomenon threatens the long-term development of such models, and some techniques have been proposed to mitigate the effect. == Characteristics == Shumailov et al. coined the term to describe two specific stages to the degradation of machine learning models: early model collapse and late model collapse: In early model collapse, the model begins losing information about the tails of the distribution – mostly affecting minority data. Later work highlighted that early model collapse is hard to notice, since overall performance may appear to improve, while the model loses performance on minority data. In late model collapse, the model loses a significant proportion of its performance, confusing concepts and losing most of its variance. == Mechanism == Using synthetic data as training data can lead to issues with the quality and reliability of the trained model. Model collapse occurs for three main reasons: functional approximation errors sampling errors learning errors Importantly, it happens in even the simplest of models, where not all of the error sources are present. In more complex models the errors often compound, leading to faster collapse. == Disagreement over real-world impact == Some researchers and commentators on model collapse warn that the phenomenon could fundamentally threaten future generative AI development: As AI-generated data is shared on the Internet, it will inevitably end up in future training datasets, which are often crawled from the Internet. If training on "slop" (large quantities of unlabeled synthetic data) inevitably leads to model collapse, this could therefore pose a difficult problem. However, recently, other researchers have disagreed with this argument, showing that if synthetic data accumulates alongside human-generated data, model collapse is avoided. The researchers argue that data accumulating over time is a more realistic description of reality than deleting all existing data every year, and that the real-world impact of model collapse may not be as catastrophic as feared. An alternative branch of the literature investigates the use of machine learning detectors and watermarking to identify model generated data and filter it out. == Mathematical models of the phenomenon == === 1D Gaussian model === In 2024, a first attempt has been made at illustrating collapse for the simplest possible model — a single dimensional normal distribution fit using unbiased estimators of mean and variance, computed on samples from the previous generation. To make this more precise, we say that original data follows a normal distribution X 0 ∼ N ( μ , σ 2 ) {\displaystyle X^{0}\sim {\mathcal {N}}(\mu ,\sigma ^{2})} , and we possess M 0 {\displaystyle M_{0}} samples X j 0 {\displaystyle X_{j}^{0}} for j ∈ { 1 , … , M 0 } {\displaystyle j\in {\{\,1,\dots ,M_{0}\,{}\}}} . Denoting a general sample X j i {\displaystyle X_{j}^{i}} as sample j ∈ { 1 , … , M i } {\displaystyle j\in {\{\,1,\dots ,M_{i}\,{}\}}} at generation i {\displaystyle i} , then the next generation model is estimated using the sample mean and variance: μ i + 1 = 1 M i ∑ j X j i ; σ i + 1 2 = 1 M i − 1 ∑ j ( X j i − μ i + 1 ) 2 . {\displaystyle \mu _{i+1}={\frac {1}{M_{i}}}\sum _{j}X_{j}^{i};\quad \sigma _{i+1}^{2}={\frac {1}{M_{i}-1}}\sum _{j}(X_{j}^{i}-\mu _{i+1})^{2}.} Leading to a conditionally normal next generation model X j i + 1 | μ i + 1 , σ i + 1 ∼ N ( μ i + 1 , σ i + 1 2 ) {\displaystyle X_{j}^{i+1}|\mu _{i+1},\;\sigma _{i+1}\sim {\mathcal {N}}(\mu _{i+1},\sigma _{i+1}^{2})} . In theory, this is enough to calculate the full distribution of X j i {\displaystyle X_{j}^{i}} . However, even after the first generation, the full distribution is no longer normal: It follows a variance-gamma distribution. To continue the analysis, instead of writing the probability density function at each generation, it is possible to explicitly construct them in terms of independent random variables using Cochran's theorem. To be precise, μ 1 {\displaystyle \mu _{1}} and σ 1 {\displaystyle \sigma _{1}} are independent, with μ 1 ∼ N ( μ , σ 2 M 0 ) {\displaystyle \mu _{1}\sim {\mathcal {N}}\left(\mu ,{\frac {\sigma ^{2}}{M_{0}}}\right)} and ( M 0 − 1 ) σ 1 2 ∼ σ 2 Γ ( M 0 − 1 2 , 1 2 ) {\displaystyle (M_{0}-1)\,\sigma _{1}^{2}\sim \sigma ^{2}\,\Gamma \left({\frac {M_{0}-1}{2}},{\frac {1}{2}}\right)} , following a Gamma distribution. Denoting with Z {\displaystyle Z} Gaussian random variables distributed according to N ( 0 , 1 ) {\displaystyle {\mathcal {N}}(0,1)} and with S i {\displaystyle S^{i}} random variables distributed with 1 M i − 1 − 1 Γ ( M i − 1 − 1 2 , 1 2 ) {\displaystyle {\frac {1}{M_{i-1}-1}}\Gamma \left({\frac {M_{i-1}-1}{2}},{\frac {1}{2}}\right)} , it turns out to be possible to write samples at each generation as X j 0 = μ + σ Z j 0 , {\textstyle X_{j}^{0}=\mu +\sigma Z_{j}^{0},} X j 1 = μ + σ M 0 Z 1 + σ S 1 Z j 1 , {\textstyle X_{j}^{1}=\mu +{\frac {\sigma }{\sqrt {M_{0}}}}Z^{1}+\sigma {\sqrt {S^{1}}}Z_{j}^{1},} and more generally X j n = μ + σ M 0 Z 1 + σ M 1 S 1 Z 2 + ⋯ + σ M n − 1 S 1 × ⋯ × S n − 1 Z n + σ S 1 × ⋯ × S n Z j n . {\displaystyle X_{j}^{n}=\mu +{\frac {\sigma }{\sqrt {M_{0}}}}Z^{1}+{\frac {\sigma }{\sqrt {M_{1}}}}{\sqrt {S^{1}}}Z^{2}+\dots +{\frac {\sigma }{\sqrt {M_{n-1}}}}{\sqrt {S^{1}\times \dots \times S^{n-1}}}Z^{n}+\sigma {\sqrt {S^{1}\times \dots \times S^{n}}}Z_{j}^{n}.} Note, that these are not joint distributions, as Z n {\displaystyle Z^{n}} and S n {\displaystyle S^{n}} depend directly on Z j n − 1 {\displaystyle Z_{j}^{n-1}} , but when considering X j n {\displaystyle X_{j}^{n}} on its own the formula above provides all the information about the full distribution. To analyse the model collapse, we can first calculate variance and mean of samples at generation n {\displaystyle n} . This would tell us what kind of distributions we expect to arrive at after n {\displaystyle n} generations. It is possible to find its exact value in closed form, but the mean and variance of the square root of gamma distribution are expressed in terms of gamma functions, making the result quite clunky. Following, it is possible to expand all results to second order in each of 1 / M i {\displaystyle 1/M_{i}} , assuming each sample size to be large. It is then possible to show that 1 σ 2 Var ⁡ ( X j n ) = 1 M 0 + 1 M 1 + ⋯ + 1 M n − 1 + 1 + O ( M i − 2 ) . {\displaystyle {\frac {1}{\sigma ^{2}}}\operatorname {Var} (X_{j}^{n})={\frac {1}{M_{0}}}+{\frac {1}{M_{1}}}+\dots +{\frac {1}{M_{n-1}}}+1+{\mathcal {O}}\left(M_{i}^{-2}\right).} And if all sample sizes M i = M {\displaystyle M_{i}=M} are constant, this diverges linearly as n → ∞ {\displaystyle n\to \infty } : Var ⁡ ( X j n ) = σ 2 ( 1 + n M ) ; E ( X j n ) = μ . {\displaystyle \operatorname {Var} (X_{j}^{n})=\sigma ^{2}\left(1+{\frac {n}{M}}\right);\quad \mathbb {E} (X_{j}^{n})=\mu .} This is the same scaling as for a single dimensional Gaussian random walk. However, divergence of the variance of X j n {\displaystyle X_{j}^{n}} does not directly provide any information about the corresponding estimates of μ n + 1 {\displaystyle \mu _{n+1}} and σ n + 1 {\displaystyle \sigma _{n+1}} , particularly how different they are from the original μ {\displaystyle \mu } and σ {\displaystyle \sigma } . It turns out to be possible to calculate the distance between the true distribution and the approximated distribution at step n + 1 {\displaystyle n+1} , using the Wasserstein-2 distance (which is also sometimes referred to as risk): E [ W 2 2 ( N ( μ , σ 2 ) , N ( μ n + 1 , σ n + 1 2 ) ) ] = 3 2 σ 2 ( 1 M 0 + 1 M 1 + ⋯ + 1 M n ) + O ( M i − 2 ) , {\displaystyle \mathbb {E} \left[\mathbb {W} _{2}^{2}\left({\mathcal {N}}(\mu ,\sigma ^{2}),{\mathcal {N}}(\mu _{n+1},\sigma _{n+1}^{2})\right)\right]={\frac {3}{2}}\sigma ^{2}\left({\frac {1}{M_{0}}}+{\frac {1}{M_{1}}}+\dots +{\frac {1}{M_{n}}}\right)+{\mathcal {O}}\left(M_{i}^{-2}\right),} Var ⁡ [ W 2 2 ( N ( μ , σ 2 ) , N ( μ n + 1 , σ n + 1 2 ) ) ] = 1 2 σ 4 ( 3 M 0 2 + 3 M 1 2 + ⋯ + 3 M n 2 + ∑ i ≠ j 4 M i M j ) + O ( M i − 3 ) . {\displaystyle \operatorname {Var} \left[\mathbb {W} _{2}^{2}\left({\mathcal {N}}(\mu ,\sigma ^{2}),{\mathcal {N}}(\mu _{n+1},\sigma _{n+1}^{2})\right)\right]={\frac {1}{2}}\sigma ^{4}\left({\frac {3}{M_{0}^{2}}}+{\frac {3}{M_{1}^{2}}}+\dots +{\frac {3}{M_{n}^{2}}}+\sum _{i\neq j}{\frac {4}{M_{i}M_{j}}}\right)+{\mathcal {O}}\left(M_{i}^{-3}\right).} This directly shows why model collapse occurs in this simple model. Due to errors from re-sampling the approximated distribution, each generation ends up corresponding to a

  Read more →
• 

  Stable Diffusion

  Stable Diffusion is a deep learning, text-to-image model released in 2022 based on diffusion techniques. The generative artificial intelligence technology is the premier product of Stability AI and is considered to be a part of the ongoing AI boom. It is primarily used to generate detailed images conditioned on text descriptions, though it can also be applied to other tasks such as inpainting, outpainting, and generating image-to-image translations guided by a text prompt. Its development involved researchers from the CompVis Group at LMU Munich and Runway with a computational donation from Stability and training data from non-profit organizations. Stable Diffusion is a latent diffusion model, a kind of deep generative artificial neural network. Its code and model weights have been released publicly, and an optimized version can run on most consumer hardware equipped with a modest GPU with as little as 2.4 GB VRAM. This marked a departure from previous proprietary text-to-image models such as DALL-E and Midjourney which were accessible only via cloud services. == Development == Stable Diffusion originated from a project called Latent Diffusion, developed in Germany by researchers at LMU Munich in Munich and Heidelberg University. Four of the original 5 authors (Robin Rombach, Andreas Blattmann, Patrick Esser and Dominik Lorenz) later joined Stability AI and released subsequent versions of Stable Diffusion. The technical license for the model was released by the CompVis group at LMU Munich. Development was led by Patrick Esser of Runway and Robin Rombach of CompVis, who were among the researchers who had earlier invented the latent diffusion model architecture used by Stable Diffusion. Stability AI also credited EleutherAI and LAION (a German nonprofit which assembled the dataset on which Stable Diffusion was trained) as supporters of the project. == Technology == === Architecture === Diffusion models, introduced in 2015, are trained with the objective of removing successive applications of Gaussian noise on training images, which can be thought of as a sequence of denoising autoencoders. The name diffusion is from the thermodynamic diffusion, since they were first developed with inspiration from thermodynamics. Models in Stable Diffusion series before SD 3 all used a variant of diffusion models, called latent diffusion model (LDM), developed in 2021 by the CompVis (Computer Vision & Learning) group at LMU Munich. Stable Diffusion consists of 3 parts: the variational autoencoder (VAE), U-Net, and an optional text encoder. The VAE encoder compresses the image from pixel space to a smaller dimensional latent space, capturing a more fundamental semantic meaning of the image. Gaussian noise is iteratively applied to the compressed latent representation during forward diffusion. The U-Net block, composed of a ResNet backbone, denoises the output from forward diffusion backwards to obtain a latent representation. Finally, the VAE decoder generates the final image by converting the representation back into pixel space. The denoising step can be flexibly conditioned on a string of text, an image, or another modality. The encoded conditioning data is exposed to denoising U-Nets via a cross-attention mechanism. For conditioning on text, the fixed, pretrained CLIP ViT-L/14 text encoder is used to transform text prompts to an embedding space. Researchers point to increased computational efficiency for training and generation as an advantage of LDMs. With 860 million parameters in the U-Net and 123 million in the text encoder, Stable Diffusion is considered relatively lightweight by 2022 standards, and unlike other diffusion models, it can run on consumer GPUs, and even CPU-only if using the OpenVINO version of Stable Diffusion. ==== SD XL ==== The XL version uses the same LDM architecture as previous versions, except larger: larger UNet backbone, larger cross-attention context, two text encoders instead of one, and trained on multiple aspect ratios (not just the square aspect ratio like previous versions). The SD XL Refiner, released at the same time, has the same architecture as SD XL, but it was trained for adding fine details to preexisting images via text-conditional img2img. ==== SD 3.0 ==== The 3.0 version completely changes the backbone. Not a UNet, but a Rectified Flow Transformer, which implements the rectified flow method with a Transformer. The Transformer architecture used for SD 3.0 has three "tracks", for original text encoding, transformed text encoding, and image encoding (in latent space). The transformed text encoding and image encoding are mixed during each transformer block. The architecture is named "multimodal diffusion transformer (MMDiT), where the "multimodal" means that it mixes text and image encodings inside its operations. This differs from previous versions of DiT, where the text encoding affects the image encoding, but not vice versa. === Training data === Stable Diffusion was trained on pairs of images and captions taken from LAION-5B, a publicly available dataset derived from Common Crawl data scraped from the web, where 5 billion image-text pairs were classified based on language and filtered into separate datasets by resolution, a predicted likelihood of containing a watermark, and predicted "aesthetic" score (e.g. subjective visual quality). The dataset was created by LAION, a German non-profit which receives funding from Stability AI. The Stable Diffusion model was trained on three subsets of LAION-5B: laion2B-en, laion-high-resolution, and laion-aesthetics v2 5+. A third-party analysis of the model's training data identified that out of a smaller subset of 12 million images taken from the original wider dataset used, approximately 47% of the sample size of images came from 100 different domains, with Pinterest taking up 8.5% of the subset, followed by websites such as WordPress, Blogspot, Flickr, DeviantArt and Wikimedia Commons. An investigation by Bayerischer Rundfunk showed that LAION's datasets, hosted on Hugging Face, contain large amounts of private and sensitive data. === Training procedures === The model was initially trained on the laion2B-en and laion-high-resolution subsets, with the last few rounds of training done on LAION-Aesthetics v2 5+, a subset of 600 million captioned images which the LAION-Aesthetics Predictor V2 predicted that humans would, on average, give a score of at least 5 out of 10 when asked to rate how much they liked them. The LAION-Aesthetics v2 5+ subset also excluded low-resolution images and images which LAION-5B-WatermarkDetection identified as carrying a watermark with greater than 80% probability. Final rounds of training additionally dropped 10% of text conditioning to improve Classifier-Free Diffusion Guidance. The model was trained using 256 Nvidia A100 GPUs on Amazon Web Services for a total of 150,000 GPU-hours, at a cost of $600,000. === Limitations === Stable Diffusion has issues with degradation and inaccuracies in certain scenarios. Initial releases of the model were trained on a dataset that consists of 512×512 resolution images, meaning that the quality of generated images noticeably degrades when user specifications deviate from its "expected" 512×512 resolution; the version 2.0 update of the Stable Diffusion model later introduced the ability to natively generate images at 768×768 resolution. Another challenge is in generating human limbs due to poor data quality of limbs in the LAION database. The model is insufficiently trained to replicate human limbs and faces due to the lack of representative features in the database, and prompting the model to generate images of such type can confound the model. In addition to human limbs, Stable Diffusion is unable to generate legible ambigrams and some other forms of text and typography. Stable Diffusion XL (SDXL) version 1.0, released in July 2023, introduced native 1024x1024 resolution and improved generation for limbs and text. Accessibility for individual developers can also be a problem. In order to customize the model for new use cases that are not included in the dataset, such as generating anime characters ("waifu diffusion"), new data and further training are required. Fine-tuned adaptations of Stable Diffusion created through additional retraining have been used for a variety of different use-cases, from medical imaging to algorithmically generated music. However, this fine-tuning process is sensitive to the quality of new data; low resolution images or different resolutions from the original data can not only fail to learn the new task but degrade the overall performance of the model. Even when the model is additionally trained on high quality images, it is difficult for individuals to run models in consumer electronics. For example, the training process for waifu-diffusion requires a minimum 30 GB of VRAM, which exceeds the usual resource provided in such consumer GPUs as Nvidia's GeForce 30 series, w

  Read more →