Developed at UC Berkeley, "Opinion Space" (also known as The Collective Discovery Engine) is a social media technology designed to help communities generate and exchange ideas about important issues and policies. Version 1.0 was launched on April 4, 2009, at UC Berkeley, and explored the question "Do you think legalizing marijuana is a good idea?" It has since undergone 4 different iterations, and been used in partnership with various organizations including The Occupy movement (Version 4.0, 5/24/2013) and the African Robots Network (Version 4.0, 5/25/2013). Opinion Space has also been used in collaboration with the United States State Department and the University of California's Berkeley Center for New Media (Version 2.0, 12/1/2009 and Version 3.0, 2/25/2012) to gain public perspective on foreign policy issues. Then U.S. Secretary of State Hillary Rodham Clinton explained, "Opinion Space will harness the power of connection technologies to provide a unique forum for international dialogue. This is...an opportunity to extend our engagement beyond the halls of government directly to the people of the world" (2010). The website uses data visualization and statistical analysis to present and develop public opinion and ideas. Opinion Space is a self-organizing system that uses an intuitive graphical "map" that displays patterns, trends, and insights as they emerge and employs the wisdom of crowds to identify and highlight the most insightful ideas. The system uses a game model that incorporates techniques from deliberative polling, collaborative filtering, and multidimensional visualization.
Gonioreflectometer
A gonioreflectometer is a device for measuring a bidirectional reflectance distribution function (BRDF). The device consists of a light source illuminating the material to be measured and a sensor that captures light reflected from that material. The light source should be able to illuminate and the sensor should be able to capture data from a hemisphere around the target. The hemispherical rotation dimensions of the sensor and light source are the four dimensions of the BRDF. The 'gonio' part of the word refers to the device's ability to measure at different angles. Several similar devices have been built and used to capture data for similar functions. Most of these devices use a camera instead of the light intensity-measuring sensor to capture a two-dimensional sample of the target. Examples include: a spatial gonioreflectometer for capturing the SBRDF (McAllister, 2002). a camera gantry for capturing the light field (Levoy and Hanrahan, 1996). an unnamed device for capturing the bidirectional texture function (Dana et al., 1999).
Emotion Markup Language
An Emotion Markup Language (EML or EmotionML) has first been defined by the W3C Emotion Incubator Group (EmoXG) as a general-purpose emotion annotation and representation language, which should be usable in a large variety of technological contexts where emotions need to be represented. Emotion-oriented computing (or "affective computing") is gaining importance as interactive technological systems become more sophisticated. Representing the emotional states of a user or the emotional states to be simulated by a user interface requires a suitable representation format; in this case a markup language is used. EmotionML version 1.0 was published by the group in May 2014. == Example == Here is an example of an EmotionML document describing emotions expressed in a video recording of the interaction between a teacher, Alice, and a student, Bob. == History == In 2006, a first W3C Incubator Group, the Emotion Incubator Group (EmoXG), was set up "to investigate a language to represent the emotional states of users and the emotional states simulated by user interfaces" with the final Report published on 10 July 2007. In 2007, the Emotion Markup Language Incubator Group (EmotionML XG) was set up as a follow-up to the Emotion Incubator Group, "to propose a specification draft for an Emotion Markup Language, to document it in a way accessible to non-experts, and to illustrate its use in conjunction with a number of existing markups." The final report of the Emotion Markup Language Incubator Group, Elements of an EmotionML 1.0, was published on 20 November 2008. The work then was continued in 2009 in the frame of the W3C's Multimodal Interaction Activity, with the First Public Working Draft of "Emotion Markup Language (EmotionML) 1.0" being published on 29 October 2009. The Last Call Working Draft of "Emotion Markup Language 1.0", was published on 7 April 2011. The Last Call Working Draft addressed all open issues that arose from feedback of the community on the First Call Working Draft as well as results of a workshop held in Paris in October 2010. Along with the Last Call Working Draft, a list of vocabularies for EmotionML has been published to aid developers using common vocabularies for annotating or representing emotions. Annual draft updates were published until the 1.0 version was finished in 2014. == Reasons for defining an emotion markup language == A standard for an emotion markup language would be useful for the following purposes: To enhance computer-mediated human-human or human-machine communication. Emotions are a basic part of human communication and should therefore be taken into account, e.g. in emotional Chat systems or emphatic voice boxes. This involves specification, analysis and display of emotion related states. To enhance systems' processing efficiency. Emotion and intelligence are strongly interconnected. The modeling of human emotions in computer processing can help to build more efficient systems, e.g. using emotional models for time-critical decision enforcement. To allow the analysis of non-verbal behavior, emotion, mental states that can be provided using web services to enable data collection, analysis, and reporting. Concrete examples of existing technology that could apply EmotionML include: Opinion mining / sentiment analysis in Web 2.0, to automatically track customer's attitude regarding a product across blogs; Affective monitoring, such as ambient assisted living applications, fear detection for surveillance purposes, or using wearable sensors to test customer satisfaction; Wellness technologies that provide assistance according to a person's emotional state with the goal to improve the person's well-being; Character design and control for games and virtual worlds; Building web services to capture, analysis, and report data of non-verbal behavior, emotion and mental states of an individual or group across the internet using standard web technologies such as HTML5 and JSON. Social robots, such as guide robots engaging with visitors; Expressive speech synthesis, generating synthetic speech with different emotions, such as happy or sad, friendly or apologetic; expressive synthetic speech would for example make more information available to blind and partially sighted people, and enrich their experience of the content; Emotion recognition (e.g., for spotting angry customers in speech dialog systems, to improve computer games or e-Learning applications); Support for people with disabilities, such as educational programs for people with autism. EmotionML can be used to make the emotional intent of content explicit. This would enable people with learning disabilities (such as Asperger syndrome) to realise the emotional context of the content; EmotionML can be used for media transcripts and captions. Where emotions are marked up to help deaf or hearing impaired people who cannot hear the soundtrack, more information is made available to enrich their experience of the content. The Emotion Incubator Group has listed 39 individual use cases for an Emotion markup language. A standardised way to mark up the data needed by such "emotion-oriented systems" has the potential to boost development primarily because data that was annotated in a standardised way can be interchanged between systems more easily, thereby simplifying a market for emotional databases, and the standard can be used to ease a market of providers for sub-modules of emotion processing systems, e.g. a web service for the recognition of emotion from text, speech or multi-modal input. == The challenge of defining a generally usable emotion markup language == Any attempt to standardize the description of emotions using a finite set of fixed descriptors is doomed to failure, as there is no consensus on the number of relevant emotions, on the names that should be given to them or how else best to describe them. For example, the difference between ":)" and "(:" is small, but using a standardized markup it would make one invalid. Even more basically, the list of emotion-related states that should be distinguished varies depending on the application domain and the aspect of emotions to be focused. Basically, the vocabulary needed depends on the context of use. On the other hand, the basic structure of concepts is less controversial: it is generally agreed that emotions involve triggers, appraisals, feelings, expressive behavior including physiological changes, and action tendencies; emotions in their entirety can be described in terms of categories or a small number of dimensions; emotions have an intensity, and so on. For details, see the Scientific Descriptions of Emotions in the Final Report of the Emotion Incubator Group. Given this lack of agreement on descriptors in the field, the only practical way of defining an emotion markup language is the definition of possible structural elements and to allow users to "plug in" vocabularies that they consider appropriate for their work. An additional challenge lies in the aim to provide a markup language that is generally usable. The requirements that arise from different use cases are rather different. Whereas manual annotation tends to require all the fine-grained distinctions considered in the scientific literature, automatic recognition systems can usually distinguish only a very small number of different states and affective avatars need yet another level of detail for expressing emotions in an appropriate way. For the reasons outlined here, it is clear that there is an inevitable tension between flexibility and interoperability, which need to be weighed in the formulation of an EmotionML. The guiding principle in the following specification has been to provide a choice only where it is needed, and to propose reasonable default options for every choice. == Applications and web services benefiting from an emotion markup language == There are a range of existing projects and applications to which an emotion markup language will enable the building of webservices to measure capture data of individuals non-verbal behavior, mental states, and emotions and allowing results to be reported and rendered in a standardized format using standard web technologies such as JSON and HTML5. One such project is measuring affect data across the Internet using EyesWeb.
Physical schema
A physical data model (or database design) is a representation of a data design as implemented, or intended to be implemented, in a database management system. In the lifecycle of a project it typically derives from a logical data model, though it may be reverse-engineered from a given database implementation. A complete physical data model will include all the database artifacts required to create relationships between tables or to achieve performance goals, such as indexes, constraint definitions, linking tables, partitioned tables or clusters. Analysts can usually use a physical data model to calculate storage estimates; it may include specific storage allocation details for a given database system. As of 2012 seven main databases dominate the commercial marketplace: Informix, Oracle, Postgres, SQL Server, Sybase, IBM Db2 and MySQL. Other RDBMS systems tend either to be legacy databases or used within academia such as universities or further education colleges. Physical data models for each implementation would differ significantly, not least due to underlying operating-system requirements that may sit underneath them. For example: SQL Server runs only on Microsoft Windows operating-systems (Starting with SQL Server 2017, SQL Server runs on Linux. It's the same SQL Server database engine, with many similar features and services regardless of your operating system), while Oracle and MySQL can run on Solaris, Linux and other UNIX-based operating-systems as well as on Windows. This means that the disk requirements, security requirements and many other aspects of a physical data model will be influenced by the RDBMS that a database administrator (or an organization) chooses to use. == Physical schema == Physical schema is a term used in data management to describe how data is to be represented and stored (files, indices, etc.) in secondary storage using a particular database management system (DBMS) (e.g., Oracle RDBMS, Sybase SQL Server, etc.). In the ANSI/SPARC Architecture three schema approach, the internal schema is the view of data that involved data management technology. This is as opposed to an external schema that reflects an individual's view of the data, or the conceptual schema that is the integration of a set of external schemas. The logical schema was the way data were represented to conform to the constraints of a particular approach to database management. At that time the choices were hierarchical and network. Describing the logical schema, however, still did not describe how physically data would be stored on disk drives. That is the domain of the physical schema. Now logical schemas describe data in terms of relational tables and columns, object-oriented classes, and XML tags. A single set of tables, for example, can be implemented in numerous ways, up to and including an architecture where table rows are maintained on computers in different countries.
Causal AI
Causal AI is a technique in artificial intelligence that builds a causal model and can thereby make inferences using causality rather than just correlation. One practical use for causal AI is for organisations to explain decision-making and the causes for a decision. Systems based on causal AI, by identifying the underlying web of causality for a behaviour or event, provide insights that solely predictive AI models might fail to extract from historical data. An analysis of causality may be used to supplement human decisions in situations where understanding the causes behind an outcome is necessary, such as quantifying the impact of different interventions, policy decisions or performing scenario planning. A 2024 paper from Google DeepMind demonstrated mathematically that "Any agent capable of adapting to a sufficiently large set of distributional shifts must have learned a causal model". The paper offers the interpretation that learning to generalise beyond the original training set requires learning a causal model, concluding that causal AI is necessary for artificial general intelligence. == History == The concept of causal AI and the limits of machine learning were raised by Judea Pearl, the Turing Award-winning computer scientist and philosopher, in 2018's The Book of Why: The New Science of Cause and Effect. Pearl asserted: “Machines' lack of understanding of causal relations is perhaps the biggest roadblock to giving them human-level intelligence.” In 2020, Columbia University established a Causal AI Lab under Director Elias Bareinboim. Professor Bareinboim's research focuses on causal and counterfactual inference and their applications to data-driven fields in the health and social sciences as well as artificial intelligence and machine learning. Technological research and consulting firm Gartner for the first time included causal AI in its 2022 Hype Cycle report, citing it as one of five critical technologies in accelerated AI automation. Causal AI is closely related to but distinct from fields such as causal inference, explainable AI and causal reasoning. While causal inference focuses on estimating cause-effect relationships (often from observational data), causal AI emphasises the integration of those causal models into AI systems for prediction, planning and adaptation.
Domain adaptation
Domain adaptation is a field associated with machine learning and transfer learning. It addresses the challenge of training a model on one data distribution (the source domain) and applying it to a related but different data distribution (the target domain). A common example is spam filtering, where a model trained on emails from one user (source domain) is adapted to handle emails for another user with significantly different patterns (target domain). Domain adaptation techniques can also leverage unrelated data sources to improve learning. When multiple source distributions are involved, the problem extends to multi-source domain adaptation. Domain adaptation is a specific type of transfer learning. According to the taxonomy laid out by Pan and Yang (2010), it falls into the category of transductive transfer learning. In this setting, the source and target tasks are the same (e.g., both are object recognition), but the domains differ (different marginal distributions). This distinguishes it from inductive transfer learning (where labeled data is available for the target task) and unsupervised transfer learning (where labels are unavailable in both domains). == Classification of domain adaptation problems == Domain adaptation setups are classified in two different ways: according to the distribution shift between the domains, and according to the available data from the target domain. === Distribution shifts === Common distribution shifts are classified as follows: Covariate Shift occurs when the input distributions of the source and destination change, but the relationship between inputs and labels remains unchanged. The above-mentioned spam filtering example typically falls in this category. Namely, the distributions (patterns) of emails may differ between the domains, but emails labeled as spam in the one domain should similarly be labeled in another. Prior Shift (Label Shift) occurs when the label distribution differs between the source and target datasets, while the conditional distribution of features given labels remains the same. An example is a classifier of hair color in images from Italy (source domain) and Norway (target domain). The proportions of hair colors (labels) differ, but images within classes like blond and black-haired populations remain consistent across domains. A classifier for the Norway population can exploit this prior knowledge of class proportions to improve its estimates. Concept Shift (Conditional Shift) refers to changes in the relationship between features and labels, even if the input distribution remains the same. For instance, in medical diagnosis, the same symptoms (inputs) may indicate entirely different diseases (labels) in different populations (domains). === Data available during training === Domain adaptation problems typically assume that some data from the target domain is available during training. Problems can be classified according to the type of this available data: Unsupervised: Unlabeled data from the target domain is available, but no labeled data. In the above-mentioned example of spam filtering, this corresponds to the case where emails from the target domain (user) are available, but they are not labeled as spam. Domain adaptation methods can benefit from such unlabeled data, by comparing its distribution (patterns) with the labeled source domain data. Semi-supervised: Most data that is available from the target domain is unlabelled, but some labeled data is also available. In the above-mentioned case of spam filter design, this corresponds to the case that the target user has labeled some emails as being spam or not. Supervised: All data that is available from the target domain is labeled. In this case, domain adaptation reduces to refinement of the source domain predictor. In the above-mentioned example classification of hair-color from images, this could correspond to the refinement of a network already trained on a large dataset of labeled images from Italy, using newly available labeled images from Norway. == Formalization == Let X {\displaystyle X} be the input space (or description space) and let Y {\displaystyle Y} be the output space (or label space). The objective of a machine learning algorithm is to learn a mathematical model (a hypothesis) h : X → Y {\displaystyle h:X\to Y} able to attach a label from Y {\displaystyle Y} to an example from X {\displaystyle X} . This model is learned from a learning sample S = { ( x i , y i ) ∈ ( X × Y ) } i = 1 m {\displaystyle S=\{(x_{i},y_{i})\in (X\times Y)\}_{i=1}^{m}} . Usually in supervised learning (without domain adaptation), we suppose that the examples ( x i , y i ) ∈ S {\displaystyle (x_{i},y_{i})\in S} are drawn i.i.d. from a distribution D S {\displaystyle D_{S}} of support X × Y {\displaystyle X\times Y} (unknown and fixed). The objective is then to learn h {\displaystyle h} (from S {\displaystyle S} ) such that it commits the least error possible for labelling new examples coming from the distribution D S {\displaystyle D_{S}} . The main difference between supervised learning and domain adaptation is that in the latter situation we study two different (but related) distributions D S {\displaystyle D_{S}} and D T {\displaystyle D_{T}} on X × Y {\displaystyle X\times Y} . The domain adaptation task then consists of the transfer of knowledge from the source domain D S {\displaystyle D_{S}} to the target one D T {\displaystyle D_{T}} . The goal is then to learn h {\displaystyle h} (from labeled or unlabelled samples coming from the two domains) such that it commits as little error as possible on the target domain D T {\displaystyle D_{T}} . The major issue is the following: if a model is learned from a source domain, what is its capacity to correctly label data coming from the target domain? == Four algorithmic principles == === Reweighting algorithms === The objective is to reweight the source labeled sample such that it "looks like" the target sample (in terms of the error measure considered). === Iterative algorithms === A method for adapting consists in iteratively "auto-labeling" the target examples. The principle is simple: a model h {\displaystyle h} is learned from the labeled examples; h {\displaystyle h} automatically labels some target examples; a new model is learned from the new labeled examples. Note that there exist other iterative approaches, but they usually need target labeled examples. === Search of a common representation space === The goal is to find or construct a common representation space for the two domains. The objective is to obtain a space in which the domains are close to each other while keeping good performances on the source labeling task. This can be achieved through the use of Adversarial machine learning techniques where feature representations from samples in different domains are encouraged to be indistinguishable. === Hierarchical Bayesian Model === The goal is to construct a Bayesian hierarchical model p ( n ) {\displaystyle p(n)} , which is essentially a factorization model for counts n {\displaystyle n} , to derive domain-dependent latent representations allowing both domain-specific and globally shared latent factors. == Software packages == Several compilations of domain adaptation and transfer learning algorithms have been implemented over the past decades: SKADA (Python) ADAPT (Python) TLlib (Python) Domain-Adaptation-Toolbox (MATLAB)
Collective operation
Collective operations are building blocks for interaction patterns, that are often used in SPMD algorithms in the parallel programming context. Hence, there is an interest in efficient realizations of these operations. A realization of the collective operations is provided by the Message Passing Interface (MPI). == Definitions == In all asymptotic runtime functions, we denote the latency α {\displaystyle \alpha } (or startup time per message, independent of message size), the communication cost per word β {\displaystyle \beta } , the number of processing units p {\displaystyle p} and the input size per node n {\displaystyle n} . In cases where we have initial messages on more than one node we assume that all local messages are of the same size. To address individual processing units we use p i ∈ { p 0 , p 1 , … , p p − 1 } {\displaystyle p_{i}\in \{p_{0},p_{1},\dots ,p_{p-1}\}} . If we do not have an equal distribution, i.e. node p i {\displaystyle p_{i}} has a message of size n i {\displaystyle n_{i}} , we get an upper bound for the runtime by setting n = max ( n 0 , n 1 , … , n p − 1 ) {\displaystyle n=\max(n_{0},n_{1},\dots ,n_{p-1})} . A distributed memory model is assumed. The concepts are similar for the shared memory model. However, shared memory systems can provide hardware support for some operations like broadcast (§ Broadcast) for example, which allows convenient concurrent read. Thus, new algorithmic possibilities can become available. == Broadcast == The broadcast pattern is used to distribute data from one processing unit to all processing units, which is often needed in SPMD parallel programs to dispense input or global values. Broadcast can be interpreted as an inverse version of the reduce pattern (§ Reduce). Initially only root r {\displaystyle r} with i d {\displaystyle id} 0 {\displaystyle 0} stores message m {\displaystyle m} . During broadcast m {\displaystyle m} is sent to the remaining processing units, so that eventually m {\displaystyle m} is available to all processing units. Since an implementation by means of a sequential for-loop with p − 1 {\displaystyle p-1} iterations becomes a bottleneck, divide-and-conquer approaches are common. One possibility is to utilize a binomial tree structure with the requirement that p {\displaystyle p} has to be a power of two. When a processing unit is responsible for sending m {\displaystyle m} to processing units i . . j {\displaystyle i..j} , it sends m {\displaystyle m} to processing unit ⌈ ( i + j ) / 2 ⌉ {\displaystyle \left\lceil (i+j)/2\right\rceil } and delegates responsibility for the processing units ⌈ ( i + j ) / 2 ⌉ . . j {\displaystyle \left\lceil (i+j)/2\right\rceil ..j} to it, while its own responsibility is cut down to i . . ⌈ ( i + j ) / 2 ⌉ − 1 {\displaystyle i..\left\lceil (i+j)/2\right\rceil -1} . Binomial trees have a problem with long messages m {\displaystyle m} . The receiving unit of m {\displaystyle m} can only propagate the message to other units, after it received the whole message. In the meantime, the communication network is not utilized. Therefore pipelining on binary trees is used, where m {\displaystyle m} is split into an array of k {\displaystyle k} packets of size ⌈ n / k ⌉ {\displaystyle \left\lceil n/k\right\rceil } . The packets are then broadcast one after another, so that data is distributed fast in the communication network. Pipelined broadcast on balanced binary tree is possible in O ( α log p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} , whereas for the non-pipelined case it takes O ( ( α + β n ) log p ) {\displaystyle {\mathcal {O}}((\alpha +\beta n)\log p)} cost. == Reduce == The reduce pattern is used to collect data or partial results from different processing units and to combine them into a global result by a chosen operator. Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} initially. All m i {\displaystyle m_{i}} are aggregated by ⊗ {\displaystyle \otimes } and the result is eventually stored on p 0 {\displaystyle p_{0}} . The reduction operator ⊗ {\displaystyle \otimes } must be associative at least. Some algorithms require a commutative operator with a neutral element. Operators like s u m {\displaystyle sum} , m i n {\displaystyle min} , m a x {\displaystyle max} are common. Implementation considerations are similar to broadcast (§ Broadcast). For pipelining on binary trees the message must be representable as a vector of smaller object for component-wise reduction. Pipelined reduce on a balanced binary tree is possible in O ( α log p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} . == All-Reduce == The all-reduce pattern (also called allreduce) is used if the result of a reduce operation (§ Reduce) must be distributed to all processing units. Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} initially. All m i {\displaystyle m_{i}} are aggregated by an operator ⊗ {\displaystyle \otimes } and the result is eventually stored on all p i {\displaystyle p_{i}} . Analog to the reduce operation, the operator ⊗ {\displaystyle \otimes } must be at least associative. All-reduce can be interpreted as a reduce operation with a subsequent broadcast (§ Broadcast). For long messages a corresponding implementation is suitable, whereas for short messages, the latency can be reduced by using a hypercube (Hypercube (communication pattern) § All-Gather/ All-Reduce) topology, if p {\displaystyle p} is a power of two. All-reduce can also be implemented with a butterfly algorithm and achieve optimal latency and bandwidth. All-reduce is possible in O ( α log p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} , since reduce and broadcast are possible in O ( α log p + β n ) {\displaystyle {\mathcal {O}}(\alpha \log p+\beta n)} with pipelining on balanced binary trees. All-reduce implemented with a butterfly algorithm achieves the same asymptotic runtime. == Prefix-Sum/Scan == The prefix-sum or scan operation is used to collect data or partial results from different processing units and to compute intermediate results by an operator, which are stored on those processing units. It can be seen as a generalization of the reduce operation (§ Reduce). Given p {\displaystyle p} processing units, message m i {\displaystyle m_{i}} is on processing unit p i {\displaystyle p_{i}} . The operator ⊗ {\displaystyle \otimes } must be at least associative, whereas some algorithms require also a commutative operator and a neutral element. Common operators are s u m {\displaystyle sum} , m i n {\displaystyle min} and m a x {\displaystyle max} . Eventually processing unit p i {\displaystyle p_{i}} stores the prefix sum ⊗ i ′ <= i {\displaystyle \otimes _{i'<=i}} m i ′ {\displaystyle m_{i'}} . In the case of the so-called exclusive prefix sum, processing unit p i {\displaystyle p_{i}} stores the prefix sum ⊗ i ′ < i {\displaystyle \otimes _{i'