The roots of the development of artificial intelligence in the People's Republic of China started in the late 1970s following Deng Xiaoping's reform and opening up emphasizing science and technology as the country's primary productive force. The initial stages of China's AI development were slow and encountered significant challenges due to lack of resources and talent. At the beginning China was behind most Western countries in terms of AI development. A majority of the research was led by scientists who had received higher education abroad. Since 2006, the Chinese government has steadily developed a national agenda for artificial intelligence development and emerged as one of the leading nations in artificial intelligence research and development. In 2016, the Chinese Communist Party (CCP) released its 13th Five-Year Plan in which it aimed to become a global AI leader by 2030. As of 2025, China is considered to be a world leader in AI technology along with the United States. The State Council has a list of "national AI teams" including fifteen China-based companies, including Baidu, Tencent, Alibaba, SenseTime, and iFlytek. Each company should lead the development of a designated specialized AI sector in China, such as facial recognition, software/hardware, and speech recognition. China's rapid AI development has significantly impacted Chinese society in many areas, including the socio-economic, military, intelligence, and political spheres. Agriculture, transportation, accommodation and food services, and manufacturing are the top industries that would be the most impacted by further AI deployment. The private sector, university laboratories, and the military are working collaboratively in many aspects as there are few current existing boundaries. In 2021, China published the Data Security Law of the People's Republic of China, its first national law addressing AI-related ethical concerns. In October 2022, the United States federal government announced a series of export controls and trade restrictions intended to restrict China's access to advanced computer chips for AI applications. In 2023, the Cyberspace Administration of China issued guidelines requiring that AI content upholds the ideology of the CCP including Core Socialist Values, avoids discrimination, respects intellectual property rights, and safeguards user data. In 2025, the Chinese government issued a document regarding training data, requiring companies to use as little as data deemed "unsafe" as possible, as well as requiring companies to test models regularly. Concerns have been raised about the effects of the Chinese government's censorship regime on the development of generative artificial intelligence and long-term talent acquisition with state of the country's demographics. Others have noted that official notions of AI safety require following the priorities of the CCP and are antithetical to standards in democratic societies and raised concerns about the extension of China's system of mass surveillance and censorship abroad. == History == The Chinese term for artificial intelligence (réngōngzhìnéng 人工智能) connotes "humanmade" intelligence. The term developed as mid-20th century localisation of the Japanese term jinko chino. The research and development of artificial intelligence in China started in the 1980s, with the announcement by Deng Xiaoping of the importance of science and technology for China's economic growth. === Late 1970s to early 2010s === Chinese artificial intelligence research and development began in late 1970s after Deng Xiaoping's reform and opening up. China's first national conference on AI occurred in 1979. Academic journals in the late 1970s began publishing literature reviews of Western research on AI topics. In the 1980s, a group of Chinese scientists launched AI research led by Qian Xuesen and Wu Wenjun. However, during the time, China's society still had a generally conservative view towards AI. In the early 1980s, Science Press published translated versions of Western textbooks such as Patrick Winston's Artificial Intelligence and Nils John Nilsson's Principles of Artificial Intelligence. In 1980, a journal of the Chinese Academy of Sciences convened its first annual National Symposium on Artificial Intelligence, which included national and international scholars like Herbert A. Simon. The Chinese Association for Artificial Intelligence (CAAI) was founded in September 1981 and was authorized by the Ministry of Civil Affairs. CAAI has continued to be the largest AI association in China as of 2025. In 1982, CAAI began publishing the Artificial Intelligence Journal, which published early AI research by Chinese academics. In the 1980s, Chinese research on AI was influenced by the field of cybernetics, particularly the work of Norbert Weiner and his text Cybernetics: Or Control and Communication in the Animal and the Machine. Chinese researchers at the time sought to situate AI as part of a broader "Intelligence Science" field which would include disciplines like mathematics, computer science, cognitive science, social sciences, and philosophy. In 1987, Tsinghua University began a research publication on AI. Beginning in 1993, smart automation and intelligence have been part of China's national technology plan. Since the 2000s, the Chinese government has further expanded its research and development funds for AI and the number of government-sponsored research projects has dramatically increased. In 2006, China announced a policy priority for the development of artificial intelligence, which was included in the National Medium and Long Term Plan for the Development of Science and Technology (2006–2020), released by the State Council. In the same year, artificial intelligence was also mentioned in the 11th Five-Year Plan. In 2011, the Association for the Advancement of Artificial Intelligence (AAAI) established a branch in Beijing, China. At same year, the Wu Wenjun Artificial Intelligence Science and Technology Award was founded in honor of Chinese mathematician Wu Wenjun, and it became the highest award for Chinese achievements in the field of artificial intelligence. The first award ceremony was held on May 14, 2012. In 2013, the International Joint Conferences on Artificial Intelligence (IJCAI) was held in Beijing, marking the first time the conference was held in China. This event coincided with the Chinese government's announcement of the "Chinese Intelligence Year," a significant milestone in China's development of artificial intelligence. === Late 2010s to early 2020s === AI became a major issue of commercial, public, and political focus in China in the latter half of the 2010s. Various interpretations of the primary cause for this increased focus exist, with some analyses focusing on the 2016 Go match between Google's AlphaGo and Lee Sedol, others emphasising the U.S. increasing trade restrictions on China's technology industries and the desire to achieve national technological self-sufficiency. The State Council of China issued "A Next Generation Artificial Intelligence Development Plan" (State Council Document [2017] No. 35) on 20 July 2017. In the document, the CCP Central Committee and the State Council urged governing bodies in China to promote the development of artificial intelligence. Specifically, the plan described AI as a strategic technology that has become a "focus of international competition".:2 The document urged significant investment in a number of strategic areas related to AI and called for close cooperation between the state and private sectors. It set the goal of China becoming the preeminent country for AI research and application by 2030. During the general secretaryship of Xi Jinping, artificial intelligence has been a focus of the CCP's military-civil fusion efforts. On the occasion of Xi's speech at the first plenary meeting of the Central Military-Civil Fusion Development Committee (CMCFDC), scholars from the National Defense University wrote in the PLA Daily that the "transferability of social resources" between economic and military ends is an essential component to being a great power. During the Two Sessions 2017,"artificial intelligence plus" was proposed to be elevated to a strategic level. The same year witnessed the emergence of multiple application-level usages in the medical field according to reports. In 2018, Xinhua News Agency, in partnership with Tencent's subsidiary Sogou, launched its first artificial intelligence-generated news anchor. In 2018, the State Council budgeted $2.1 billion for an AI industrial park in Mentougou district. In order to achieve this the State Council stated the need for massive talent acquisition, theoretical and practical developments, as well as public and private investments. Some of the stated motivations that the State Council gave for pursuing its AI strategy include the potential of artificial intelligence for industrial transformation, better social
Tessellation (computer graphics)
In computer graphics, tessellation is the dividing of datasets of polygons (sometimes called vertex sets) presenting objects in a scene into suitable structures for rendering. Especially for real-time rendering, data is tessellated into triangles, for example in OpenGL 4.0 and Direct3D 11. == In graphics rendering == A key advantage of tessellation for realtime graphics is that it allows detail to be dynamically added and subtracted from a 3D polygon mesh and its silhouette edges based on control parameters (often camera distance). In previously leading realtime techniques such as parallax mapping and bump mapping, surface details could be simulated at the pixel level, but silhouette edge detail was fundamentally limited by the quality of the original dataset. In Direct3D 11 pipeline (a part of DirectX 11), the graphics primitive is the patch. The tessellator generates a triangle-based tessellation of the patch according to tessellation parameters such as the TessFactor, which controls the degree of fineness of the mesh. The tessellation, along with shaders such as a Phong shader, allows for producing smoother surfaces than would be generated by the original mesh. By offloading the tessellation process onto the GPU hardware, smoothing can be performed in real time. Tessellation can also be used for implementing subdivision surfaces, level of detail scaling and fine displacement mapping. OpenGL 4.0 uses a similar pipeline, where tessellation into triangles is controlled by the Tessellation Control Shader and a set of four tessellation parameters. == In computer-aided design == In computer-aided design the constructed design is represented by a boundary representation topological model, where analytical 3D surfaces and curves, limited to faces, edges, and vertices, constitute a continuous boundary of a 3D body. Arbitrary 3D bodies are often too complicated to analyze directly. So they are approximated (tessellated) with a mesh of small, easy-to-analyze pieces of 3D volume—usually either irregular tetrahedra, or irregular hexahedra. The mesh is used for finite element analysis. The mesh of a surface is usually generated per individual faces and edges (approximated to polylines) so that original limit vertices are included into mesh. To ensure that approximation of the original surface suits the needs of further processing, three basic parameters are usually defined for the surface mesh generator: The maximum allowed distance between the planar approximation polygon and the surface (known as "sag"). This parameter ensures that mesh is similar enough to the original analytical surface (or the polyline is similar to the original curve). The maximum allowed size of the approximation polygon (for triangulations it can be maximum allowed length of triangle sides). This parameter ensures enough detail for further analysis. The maximum allowed angle between two adjacent approximation polygons (on the same face). This parameter ensures that even very small humps or hollows that can have significant effect to analysis will not disappear in mesh. An algorithm generating a mesh is typically controlled by the above three and other parameters. Some types of computer analysis of a constructed design require an adaptive mesh refinement, which is a mesh made finer (using stronger parameters) in regions where the analysis needs more detail.
Lai–Robbins lower bound
The Lai–Robbins lower bound gives an asymptotic lower bound on the regret that any uniformly good algorithm must incur in the stochastic multi-armed bandit problem. The original result was proved by Tze Leung Lai and Herbert Robbins in 1985 for parametric exponential families. Later work extended the statement to more general classes of distributions. == Multi-armed bandit problem == The multi-armed bandit problem (MAB) is a sequential game in which the player must trade off exploration (to learn) and exploitation (to earn). The player chooses among K {\displaystyle K} actions (arms) with unknown distributions ν = ( ν 1 , … , ν K ) {\displaystyle \nu =(\nu _{1},\dots ,\nu _{K})} . The player is assumed to know a class of distributions D {\displaystyle {\mathcal {D}}} such that for every k {\displaystyle k} one has ν k ∈ D {\displaystyle \nu _{k}\in {\mathcal {D}}} (for example, D {\displaystyle {\mathcal {D}}} may be the family of Gaussian or Bernoulli distributions). At each round t = 1 , … , T {\displaystyle t=1,\dots ,T} the player selects (pulls) an arm a t {\displaystyle a_{t}} and observes a reward X t ∼ ν a t {\displaystyle X_{t}\sim \nu _{a_{t}}} . We denote N a ( t ) := ∑ s = 1 t 1 { a s = a } {\displaystyle N_{a}(t):=\sum _{s=1}^{t}\mathbf {1} _{\{a_{s}=a\}}} the number of times arm a {\displaystyle a} has been pulled in the first t {\displaystyle t} rounds, μ ( ν ) := ( μ 1 , … , μ K ) {\displaystyle \mu (\nu ):=(\mu _{1},\dots ,\mu _{K})} the vector of arm means, where μ k = E X ∼ ν k [ X ] {\displaystyle \mu _{k}=\mathbb {E} _{X\sim \nu _{k}}[X]} , μ ∗ := max a μ a {\displaystyle \mu ^{}:=\max _{a}\mu _{a}} the highest mean Δ a := μ ∗ − μ a ≥ 0 {\displaystyle \Delta _{a}:=\mu ^{}-\mu _{a}\geq 0} the gap of arm a {\displaystyle a} . An arm a {\displaystyle a} with μ a = μ ∗ {\displaystyle \mu _{a}=\mu ^{}} is called an optimal arm; otherwise it is a suboptimal arm. The goal is to minimize the regret at horizon T {\displaystyle T} , defined by R T := ∑ a = 1 K Δ a E [ N a ( T ) ] . {\displaystyle R_{T}:=\sum _{a=1}^{K}\Delta _{a}\,\mathbb {E} [N_{a}(T)].} Intuitively, the regret is the (expected) total loss compared to always playing an optimal arm: regret = ∑ a ( cost of playing a ) × ( times a is played ) . {\displaystyle {\text{regret}}=\sum _{a}\ ({\text{cost of playing }}a)\times ({\text{times }}a{\text{ is played}}).} An MAB algorithm is a (possibly randomized) policy that, at each round t {\displaystyle t} , choose an arm a_t by using the observations received from previous turns. === Intuitive example === Suppose a farmer must choose, each year, one of K {\displaystyle K} seed varieties to plant. Each variety k {\displaystyle k} has an unknown average yield μ k {\displaystyle \mu _{k}} . If the farmer knew the best variety (with mean μ ∗ {\displaystyle \mu ^{}} ) he would plant it every year; in reality he must try varieties to learn which is best. The cumulative regret after T {\displaystyle T} years measures the total expected loss in yield due to imperfect knowledge. Remarks The model above is the stochastic MAB; there also exist adversarial variants. One may consider a fixed-horizon setting (known T {\displaystyle T} ) or an anytime setting (unknown T {\displaystyle T} ). == Lai–Robbins lower bound == The theorem gives the right amount of time we should pull a suboptimal arm k {\displaystyle k} to distinguish whether we are in the instance with ν k {\displaystyle \nu _{k}} or with ν ~ k {\displaystyle {\tilde {\nu }}_{k}} where ν ~ k {\displaystyle {\tilde {\nu }}_{k}} is such that μ ~ k > μ ∗ {\displaystyle {\tilde {\mu }}_{k}>\mu ^{}} . Knowning a lower bound on the number of pull of every suboptimal arm gives a lower bound on the regret as only suboptimal arms contribute to the regret. Before stating the formal theorem we need to define what is a consistent algorithm. === Consistency (uniformly good algorithms) === Let D {\displaystyle {\mathcal {D}}} be a class of probability distributions and consider K {\displaystyle K} arms with reward distributions ν = ( ν 1 , … , ν K ) ∈ D K {\displaystyle \nu =(\nu _{1},\dots ,\nu _{K})\in {\mathcal {D}}^{K}} . An algorithm is said to be consistent (also called uniformly good) on D K {\displaystyle {\mathcal {D}}^{K}} if, for every instance ν ∈ D K {\displaystyle \nu \in {\mathcal {D}}^{K}} , the expected regret R T ( ν ) {\displaystyle R_{T}(\nu )} grows subpolynomially: ∀ α > 0 , R T ( ν ) = o ( T α ) as T → ∞ {\displaystyle \forall \alpha >0,\qquad R_{T}(\nu )=o(T^{\alpha })\quad {\text{as }}T\to \infty } This assumption excludes algorithms that perform well on some instances but incur linear regret on others. === Formal lower bound === For any suboptimal arm a {\displaystyle a} . For a distribution ν a ∈ D {\displaystyle \nu _{a}\in {\mathcal {D}}} and a threshold x {\displaystyle x} , define K inf ( ν a , x , D ) := inf { KL ( ν a , ν ′ ) : ν ′ ∈ D , μ ′ > x } {\displaystyle {\mathcal {K}}_{\inf }(\nu _{a},x,{\mathcal {D}}):=\inf {\Bigl \{}\operatorname {KL} (\nu _{a},\nu '):\nu '\in {\mathcal {D}},\ \mu '>x{\Bigr \}}} where KL ( ⋅ , ⋅ ) {\displaystyle \operatorname {KL} (\cdot ,\cdot )} denotes the Kullback-Leibler divergence. Then, for any algorithm consistent on D K {\displaystyle {\mathcal {D}}^{K}} and for every instance ν ∈ D K {\displaystyle \nu \in {\mathcal {D}}^{K}} , every suboptimal arm a {\displaystyle a} satisfies E ν [ N a ( T ) ] ≥ ln T K inf ( ν a , μ ∗ , D ) + o ( ln T ) {\displaystyle \mathbb {E} _{\nu }[N_{a}(T)]\geq {\frac {\ln T}{{\mathcal {K}}_{\inf }(\nu _{a},\mu ^{},{\mathcal {D}})}}+o(\ln T)} Consequently, the regret satisfies R T ( ν ) ≥ ( ∑ a : μ a < μ ∗ Δ a K inf ( ν a , μ ∗ , D ) ) ln T + o ( ln T ) {\displaystyle R_{T}(\nu )\geq \left(\sum _{a:\,\mu _{a}<\mu ^{}}{\frac {\Delta _{a}}{{\mathcal {K}}_{\inf }(\nu _{a},\mu ^{},{\mathcal {D}})}}\right)\ln T+o(\ln T)} The original 1985 paper established this result for exponential families; later work showed that the bound holds under much weaker assumptions on D {\displaystyle {\mathcal {D}}} . === Intuition === Consistency imposes that, for every ν {\displaystyle \nu } , the number of pulls of an optimal arm must be large. This means that μ ∗ {\displaystyle \mu ^{}} is estimated very accurately. The goal is to determine, for a suboptimal arm k {\displaystyle k} , how many samples are needed to be confident, with the appropriate level of confidence, that μ k < μ ∗ {\displaystyle \mu _{k}<\mu ^{}} . To do so, we use what is called the most confusing instance: an instance close to ν {\displaystyle \nu } such that arm k {\displaystyle k} is optimal. We define it as ν ~ {\displaystyle {\tilde {\nu }}} such that, for all a ≠ k {\displaystyle a\neq k} , ν ~ a = ν a {\displaystyle {\tilde {\nu }}_{a}=\nu _{a}} , and ν ~ k {\displaystyle {\tilde {\nu }}_{k}} is chosen so that μ ~ k > μ ∗ {\displaystyle {\tilde {\mu }}_{k}>\mu ^{}} . The objective is to determine how many samples of arm k {\displaystyle k} are required to distinguish whether we are in the instance with ν k {\displaystyle \nu _{k}} or with ν ~ k {\displaystyle {\tilde {\nu }}_{k}} in terms of KL {\displaystyle \operatorname {KL} } distance. == Algorithms achieving the Lai–Robbins lower bound == Several algorithms are known to achieve the Lai–Robbins asymptotic lower bound under specific assumptions on the reward distribution class D {\displaystyle {\mathcal {D}}} . The following list summarizes a non-exhaustive list of algorithms matching the lower bound. == Extension to other problems == === Structured bandit === A more complexe is structured bandit where we know that the mean of each arm is in a set with some restriction. In this case we can prove a smaller lower bound that use the knowledge of this set. === Best arm identification (BAI) === A similar result has been proved for best arm identification, which is the same game except that, instead of minimizing the regret, the goal is to identify the best arm with probability 1 − δ {\displaystyle 1-\delta } using as few rounds as possible. === Reinforcement Learning (RL) === Similar results have been proved for regret minimization in average-reward reinforcement learning. The order is also ln T {\displaystyle \ln T} , with a constant that depends on the problem.
Navigational database
A navigational database is a type of database in which records or objects are found primarily by following references from other objects. The term was popularized by the title of Charles Bachman's 1973 Turing Award paper, The Programmer as Navigator. This paper emphasized the fact that the new disk-based database systems allowed the programmer to choose arbitrary navigational routes following relationships from record to record, contrasting this with the constraints of earlier magnetic-tape and punched card systems where data access was strictly sequential. One of the earliest navigational databases was Integrated Data Store (IDS), which was developed by Bachman for General Electric in the 1960s. IDS became the basis for the CODASYL database model in 1969. Although Bachman described the concept of navigation in abstract terms, the idea of navigational access came to be associated strongly with the procedural design of the CODASYL Data Manipulation Language. Writing in 1982, for example, Tsichritzis and Lochovsky state that "The notion of currency is central to the concept of navigation." By the notion of currency, they refer to the idea that a program maintains (explicitly or implicitly) a current position in any sequence of records that it is processing, and that operations such as GET NEXT and GET PRIOR retrieve records relative to this current position, while also changing the current position to the record that is retrieved. Navigational database programming thus came to be seen as intrinsically procedural; and moreover to depend on the maintenance of an implicit set of global variables (currency indicators) holding the current state. As such, the approach was seen as diametrically opposed to the declarative programming style used by the relational model. The declarative nature of relational languages such as SQL offered better programmer productivity and a higher level of data independence (that is, the ability of programs to continue working as the database structure evolves.) Navigational interfaces, as a result, were gradually eclipsed during the 1980s by declarative query languages. During the 1990s it started becoming clear that for certain applications handling complex data (for example, spatial databases and engineering databases), the relational calculus had limitations. At that time, a reappraisal of the entire database market began, with several companies describing the new systems using the marketing term NoSQL. Many of these systems introduced data manipulation languages which, while far removed from the CODASYL DML with its currency indicators, could be understood as implementing Bachman's "navigational" vision. Some of these languages are procedural; others (such as XPath) are entirely declarative. Offshoots of the navigational concept, such as the graph database, found new uses in modern transaction processing workloads. == Description == Navigational access is traditionally associated with the network model and hierarchical model of database, and conventionally describes data manipulation APIs in which records (or objects) are processed one at a time, iteratively. The essential characteristic as described by Bachman, however, is finding records by virtue of their relationship to other records: so an interface can still be navigational if it has set-oriented features. From this viewpoint, the key difference between navigational data manipulation languages and relational languages is the use of explicit named relationships rather than value-based joins: for department with name="Sales", find all employees in set department-employees versus find employees, departments where employee.department-code = department.code and department.name="Sales". In practice, however, most navigational APIs have been procedural: the above query would be executed using procedural logic along the lines of the following pseudo-code: On this viewpoint, the key difference between navigational APIs and the relational model (implemented in relational databases) is that relational APIs use "declarative" or logic programming techniques that ask the system what to fetch, while navigational APIs instruct the system in a sequence of steps how to reach the required records. Most criticisms of navigational APIs fall into one of two categories: Usability: application code quickly becomes unreadable and difficult to debug Data independence: application code needs to change whenever the data structure changes For many years the primary defence of navigational APIs was performance. Database systems that support navigational APIs often use internal storage structures that contain physical links or pointers from one record to another. While such structures may allow very efficient navigation, they have disadvantages because it becomes difficult to reorganize the physical placement of data. It is quite possible to implement navigational APIs without low-level pointer chasing (Bachman's paper envisaged logical relationships being implemented just as in relational systems, using primary keys and foreign keys), so the two ideas should not be conflated. But without the performance benefits of low-level pointers, navigational APIs become harder to justify. Hierarchical models often construct primary keys for records by concatenating the keys that appear at each level in the hierarchy. Such composite identifiers are found in computer file names (/usr/david/docs/index.txt), in URIs, in the Dewey decimal system, and for that matter in postal addresses. Such a composite key can be considered as representing a navigational path to a record; but equally, it can be considered as a simple primary key allowing associative access. As relational systems came to prominence in the 1980s, navigational APIs (and in particular, procedural APIs) were criticized and fell out of favour. The 1990s, however, brought a new wave of object-oriented databases that often provided both declarative and procedural interfaces. One explanation for this is that they were often used to represent graph-structured information (for example spatial data and engineering data) where access is inherently recursive: the mathematics originally underpinning SQL (specifically, first-order predicate calculus) does not have sufficient power to support recursive queries, even those as simple as a transitive closure. More recent SQL implementations do support hierarchical and recursive queries. A current example of a popular navigational API can be found in the Document Object Model (DOM) often used in web browsers and closely associated with JavaScript. The DOM is essentially an in-memory hierarchical database with an API that is both procedural and navigational. By contrast, the same data (XML or HTML) can be accessed using XPath, which can be categorized as declarative and navigational: data is accessed by following relationships, but the calling program does not issue a sequence of instructions to be followed in order. Languages such as SPARQL used to retrieve Linked Data from the Semantic Web are also simultaneously declarative and navigational. == Examples == IBM Information Management System IDMS
Small data
Small data is data that is 'small' enough for human comprehension. It is data in a volume and format that makes it accessible, informative and actionable. The term "big data" is about machines and "small data" is about people. This is to say that eyewitness observations or five pieces of related data could be small data. Small data is what we used to think of as data. The only way to comprehend Big data is to reduce the data into small, visually-appealing objects representing various aspects of large data sets (such as histogram, charts, and scatter plots). Big Data is all about finding correlations, but Small Data is all about finding the causation, the reason why. A formal definition of small data has been proposed by Allen Bonde, former vice-president of Innovation at Actuate - now part of OpenText: "Small data connects people with timely, meaningful insights (derived from big data and/or “local” sources), organized and packaged – often visually – to be accessible, understandable, and actionable for everyday tasks." Another definition of small data is: The small set of specific attributes produced by the Internet of Things. These are typically a small set of sensor data such as temperature, wind speed, vibration and status. It was estimated (2016) that “If one takes the top 100 biggest innovations of our time, perhaps around 60% to 65% percent are really based on Small Data.” as Martin Lindstrom puts it. Small data includes everything from Snapchat to simple objects such as the post-it note. Lindstrom believes we become so focused on Big-Data that we tend to forget about more basic concepts and creativity. Lindstrom defines Small Data "as seemingly insignificant observations you identify in consumers’ homes, is everything from how you place your shoes on how you hang your paintings". He thus considers that one should perfectly master the basic (Small Data) in order to mine and find correlations. == Academic Recognition and Methodology == The growing significance of "small data" as a distinct field of inquiry was highlighted by the 2024 Thematic Einstein Semester (TES) on Small Data Analysis, hosted by the Berlin Mathematics Research Center MATH+. A central focus of this semester was the transition from theoretical analysis to practical decision-making. Because small data sets are primarily used to drive specific actions, the presentation of results becomes an essential methodological step. The semester’s findings emphasized that while small data may lack volume, it often contains a high density of "many possible interpretations." Consequently, the final conference of the TES was structured around the pillars of interpretation, explanation, and knowledge gain. Participants sought to develop new mathematical and methodical representations that could accurately depict this wealth of interpretative possibilities. This work underscores that analyzing small data is not purely a computational task; it requires a robust interface between mathematics and diverse disciplines to ensure that insights are both contextually grounded and scientifically rigorous. == Uses in business == === Marketing === Bonde has written about the topic for Forbes, Direct Marketing News, CMO.com and other publications. According to Martin Lindstrom, in his book, Small Data: "{In customer research, small data is} Seemingly insignificant behavioural observations containing very specific attributes pointing towards an unmet customer need. Small data is the foundation for breakthrough ideas or completely new ways to turnaround brands." His approach is based on the combination of the observation of small samples with intuition. Marketers can obtain market insights from gathering Small Data by engaging with and observing people in their own environments. In comparison to Big Data, Small Data has the power to trigger emotions and to provide insights into the reasons behind the behaviours of customers. It may uncover detailed information on a person's extroversion or introversion, self-confidence, whether one is having problems in his/her relationship, etc. According to Lindstrom, relationships among people and customer segments are organized around four criteria: Climate: It reveals for example how a person's environment affects their diet. Rulership: The power or government in charge Religion: The prevalence of religion in a country, depending on its influence, indicates whether a person's decision making process is impacted by their belief system. Tradition: Cultural norms influence people's behaviors and interactions. Many companies underestimate the power of Small Data, using samples of millions of consumers instead of recognizing the value of closely observing small samples in their market research. In his book, Lindstrom defines "7Cs", which companies should consider in the attempt to derive meaningful customer insights and market trends through small data from their customers: Collecting: Understanding the manner in which observations are translated inside a home. Clues: Uncovering other distinctive emotional reflections that can be observed. Connecting: Identifying the consequences of emotional behaviour. Causation: Understanding what emotions are being evoked. Correlation: Identifying the initial date of appearance of the behaviour or emotion. Compensation: Identifying the unmet or unfulfilled desire. Concept: Defining the “big idea” compensation for the identified consumer need. Some of Lindstrom's clients such as Lowes Foods looked at data in a different way and actually chose to live with the customer. “As you enter their store, they have now created an amazing community where every staff member acts in a character mood, based on Small Data”. The supermarket made everything it can to make the customer feel at home. All the behaviours of employees are inspired by customer feedbacks gathered from interviews directly done at customer’s home. === Healthcare === Researchers at Cornell University started developing applications to monitor health problems in patients, based on small data. This is an initiative of Cornell's Small Data Lab, in close cooperation with Weill Cornell Medicine College, led by Deborah Estrin. The Small Data Lab developed a series of apps, focusing not only on gathering data from patients' pain but also tracking habits in areas such as grocery shopping. In the case of patients with rheumatoid arthritis for example, which has flares and remissions that do not follow a particular cycle, the app gathers information passively, thus allowing to forecast when a flare might be coming up based on small changes in behaviour. Other apps developed also include monitoring online grocery shopping, to use this information from every user to adapt their groceries to the recommendations of nutritionists, or monitoring email language to identify patterns that might indicate "fluctuations in cognitive performance, fatigue, side effects of medication or poor sleep, and other conditions and treatments that are typically self-reported and self-medicated". === Postal Service === The United States Postal Service (USPS) used optical character recognition (OCR) to automatically read and process 98% of all hand-addressed mail and 99.5% of machine-printed mail. By combining this technology with its small data sample of US zip codes, the USPS can now process more than 36,000 pieces of mail per hour. === Aerospace === In 2015, Boeing established the analytics lab for aerospace data in cooperation with the Carnegie Mellon University to leverage the university's leadership in machine learning, language technologies and data analytics. One of the initiatives projects aims to by standardize maintenance logs using AI to dramatically reduce costs. Currently, there is no standardized procedure to document maintenance logs leading to small but highly unstructured data sets. As a result, it becomes highly difficult for maintenance workers to translate these variations in maintenance logs within a short period of time. However, with AI and a narrow data set of common aircraft maintenance terminology, it becomes possible to dynamically translate these logs in real time. By using AI to enhance the speed and accuracy of the airline maintenance workflow, airlines stand to save billions according to the Harvard Business Review.
Image moment
In image processing, computer vision and related fields, an image moment is a certain particular weighted average (moment) of the image pixels' intensities, or a function of such moments, usually chosen to have some attractive property or interpretation. Image moments are useful to describe objects after segmentation. Simple properties of the image which are found via image moments include area (or total intensity), its centroid, and information about its orientation. == Raw moments == For a 2D continuous function f(x,y) the moment (sometimes called "raw moment") of order (p + q) is defined as M p q = ∫ − ∞ ∞ ∫ − ∞ ∞ x p y q f ( x , y ) d x d y {\displaystyle M_{pq}=\int \limits _{-\infty }^{\infty }\int \limits _{-\infty }^{\infty }x^{p}y^{q}f(x,y)\,dx\,dy} for p,q = 0,1,2,... Adapting this to scalar (grayscale) image with pixel intensities I(x,y), raw image moments Mij are calculated by M i j = ∑ x ∑ y x i y j I ( x , y ) {\displaystyle M_{ij}=\sum _{x}\sum _{y}x^{i}y^{j}I(x,y)\,\!} In some cases, this may be calculated by considering the image as a probability density function, i.e., by dividing the above by ∑ x ∑ y I ( x , y ) {\displaystyle \sum _{x}\sum _{y}I(x,y)\,\!} A uniqueness theorem states that if f(x,y) is piecewise continuous and has nonzero values only in a finite part of the xy plane, moments of all orders exist, and the moment sequence (Mpq) is uniquely determined by f(x,y). Conversely, (Mpq) uniquely determines f(x,y). In practice, the image is summarized with functions of a few lower order moments. === Examples === Simple image properties derived via raw moments include: Area (for binary images) or sum of grey level (for greytone images): M 00 {\displaystyle M_{00}} Centroid: { x ¯ , y ¯ } = { M 10 M 00 , M 01 M 00 } {\displaystyle \{{\bar {x}},\ {\bar {y}}\}=\left\{{\frac {M_{10}}{M_{00}}},{\frac {M_{01}}{M_{00}}}\right\}} == Central moments == Central moments are defined as μ p q = ∫ − ∞ ∞ ∫ − ∞ ∞ ( x − x ¯ ) p ( y − y ¯ ) q f ( x , y ) d x d y {\displaystyle \mu _{pq}=\int \limits _{-\infty }^{\infty }\int \limits _{-\infty }^{\infty }(x-{\bar {x}})^{p}(y-{\bar {y}})^{q}f(x,y)\,dx\,dy} where x ¯ = M 10 M 00 {\displaystyle {\bar {x}}={\frac {M_{10}}{M_{00}}}} and y ¯ = M 01 M 00 {\displaystyle {\bar {y}}={\frac {M_{01}}{M_{00}}}} are the components of the centroid. If ƒ(x, y) is a digital image, then the previous equation becomes μ p q = ∑ x ∑ y ( x − x ¯ ) p ( y − y ¯ ) q f ( x , y ) {\displaystyle \mu _{pq}=\sum _{x}\sum _{y}(x-{\bar {x}})^{p}(y-{\bar {y}})^{q}f(x,y)} The central moments of order up to 3 are: μ 00 = M 00 , μ 01 = 0 , μ 10 = 0 , μ 11 = M 11 − x ¯ M 01 = M 11 − y ¯ M 10 , μ 20 = M 20 − x ¯ M 10 , μ 02 = M 02 − y ¯ M 01 , μ 21 = M 21 − 2 x ¯ M 11 − y ¯ M 20 + 2 x ¯ 2 M 01 , μ 12 = M 12 − 2 y ¯ M 11 − x ¯ M 02 + 2 y ¯ 2 M 10 , μ 30 = M 30 − 3 x ¯ M 20 + 2 x ¯ 2 M 10 , μ 03 = M 03 − 3 y ¯ M 02 + 2 y ¯ 2 M 01 . {\displaystyle {\begin{aligned}\mu _{00}&=M_{00},&\mu _{01}&=0,\\\mu _{10}&=0,&\mu _{11}&=M_{11}-{\bar {x}}M_{01}=M_{11}-{\bar {y}}M_{10},\\\mu _{20}&=M_{20}-{\bar {x}}M_{10},&\mu _{02}&=M_{02}-{\bar {y}}M_{01},\\\mu _{21}&=M_{21}-2{\bar {x}}M_{11}-{\bar {y}}M_{20}+2{\bar {x}}^{2}M_{01},&\mu _{12}&=M_{12}-2{\bar {y}}M_{11}-{\bar {x}}M_{02}+2{\bar {y}}^{2}M_{10},\\\mu _{30}&=M_{30}-3{\bar {x}}M_{20}+2{\bar {x}}^{2}M_{10},&\mu _{03}&=M_{03}-3{\bar {y}}M_{02}+2{\bar {y}}^{2}M_{01}.\end{aligned}}} It can be shown that: μ p q = ∑ m p ∑ n q ( p m ) ( q n ) ( − x ¯ ) ( p − m ) ( − y ¯ ) ( q − n ) M m n {\displaystyle \mu _{pq}=\sum _{m}^{p}\sum _{n}^{q}{p \choose m}{q \choose n}(-{\bar {x}})^{(p-m)}(-{\bar {y}})^{(q-n)}M_{mn}} Central moments are translational invariant. === Examples === Information about image orientation can be derived by first using the second order central moments to construct a covariance matrix. μ 20 ′ = μ 20 / μ 00 = M 20 / M 00 − x ¯ 2 μ 02 ′ = μ 02 / μ 00 = M 02 / M 00 − y ¯ 2 μ 11 ′ = μ 11 / μ 00 = M 11 / M 00 − x ¯ y ¯ {\displaystyle {\begin{aligned}\mu '_{20}&=\mu _{20}/\mu _{00}=M_{20}/M_{00}-{\bar {x}}^{2}\\\mu '_{02}&=\mu _{02}/\mu _{00}=M_{02}/M_{00}-{\bar {y}}^{2}\\\mu '_{11}&=\mu _{11}/\mu _{00}=M_{11}/M_{00}-{\bar {x}}{\bar {y}}\end{aligned}}} The covariance matrix of the image I ( x , y ) {\displaystyle I(x,y)} is now cov [ I ( x , y ) ] = [ μ 20 ′ μ 11 ′ μ 11 ′ μ 02 ′ ] . {\displaystyle \operatorname {cov} [I(x,y)]={\begin{bmatrix}\mu '_{20}&\mu '_{11}\\\mu '_{11}&\mu '_{02}\end{bmatrix}}.} The eigenvectors of this matrix correspond to the major and minor axes of the image intensity, so the orientation can thus be extracted from the angle of the eigenvector associated with the largest eigenvalue towards the axis closest to this eigenvector. It can be shown that this angle Θ is given by the following formula: Θ = 1 2 arctan ( 2 μ 11 ′ μ 20 ′ − μ 02 ′ ) {\displaystyle \Theta ={\frac {1}{2}}\arctan \left({\frac {2\mu '_{11}}{\mu '_{20}-\mu '_{02}}}\right)} The above formula holds as long as: μ 20 ′ − μ 02 ′ ≠ 0 {\displaystyle \mu '_{20}-\mu '_{02}\neq 0} The eigenvalues of the covariance matrix can easily be shown to be λ i = μ 20 ′ + μ 02 ′ 2 ± 4 μ ′ 11 2 + ( μ ′ 20 − μ ′ 02 ) 2 2 , {\displaystyle \lambda _{i}={\frac {\mu '_{20}+\mu '_{02}}{2}}\pm {\frac {\sqrt {4{\mu '}_{11}^{2}+({\mu '}_{20}-{\mu '}_{02})^{2}}}{2}},} and are proportional to the squared length of the eigenvector axes. The relative difference in magnitude of the eigenvalues are thus an indication of the eccentricity of the image, or how elongated it is. The eccentricity is 1 − λ 2 λ 1 . {\displaystyle {\sqrt {1-{\frac {\lambda _{2}}{\lambda _{1}}}}}.} == Moment invariants == Moments are well-known for their application in image analysis, since they can be used to derive invariants with respect to specific transformation classes. The term invariant moments is often abused in this context. However, while moment invariants are invariants that are formed from moments, the only moments that are invariants themselves are the central moments. Note that the invariants detailed below are exactly invariant only in the continuous domain. In a discrete domain, neither scaling nor rotation are well defined: a discrete image transformed in such a way is generally an approximation, and the transformation is not reversible. These invariants therefore are only approximately invariant when describing a shape in a discrete image. === Translation invariants === The central moments μi j of any order are, by construction, invariant with respect to translations. === Scale invariants === Invariants ηi j with respect to both translation and scale can be constructed from central moments by dividing through a properly scaled zero-th central moment: η i j = μ i j μ 00 ( 1 + i + j 2 ) {\displaystyle \eta _{ij}={\frac {\mu _{ij}}{\mu _{00}^{\left(1+{\frac {i+j}{2}}\right)}}}\,\!} where i + j ≥ 2. Note that translational invariance directly follows by only using central moments. === Rotation invariants === As shown in the work of Hu, invariants with respect to translation, scale, and rotation can be constructed: I 1 = η 20 + η 02 {\displaystyle I_{1}=\eta _{20}+\eta _{02}} I 2 = ( η 20 − η 02 ) 2 + 4 η 11 2 {\displaystyle I_{2}=(\eta _{20}-\eta _{02})^{2}+4\eta _{11}^{2}} I 3 = ( η 30 − 3 η 12 ) 2 + ( 3 η 21 − η 03 ) 2 {\displaystyle I_{3}=(\eta _{30}-3\eta _{12})^{2}+(3\eta _{21}-\eta _{03})^{2}} I 4 = ( η 30 + η 12 ) 2 + ( η 21 + η 03 ) 2 {\displaystyle I_{4}=(\eta _{30}+\eta _{12})^{2}+(\eta _{21}+\eta _{03})^{2}} I 5 = ( η 30 − 3 η 12 ) ( η 30 + η 12 ) [ ( η 30 + η 12 ) 2 − 3 ( η 21 + η 03 ) 2 ] + ( 3 η 21 − η 03 ) ( η 21 + η 03 ) [ 3 ( η 30 + η 12 ) 2 − ( η 21 + η 03 ) 2 ] {\displaystyle I_{5}=(\eta _{30}-3\eta _{12})(\eta _{30}+\eta _{12})[(\eta _{30}+\eta _{12})^{2}-3(\eta _{21}+\eta _{03})^{2}]+(3\eta _{21}-\eta _{03})(\eta _{21}+\eta _{03})[3(\eta _{30}+\eta _{12})^{2}-(\eta _{21}+\eta _{03})^{2}]} I 6 = ( η 20 − η 02 ) [ ( η 30 + η 12 ) 2 − ( η 21 + η 03 ) 2 ] + 4 η 11 ( η 30 + η 12 ) ( η 21 + η 03 ) {\displaystyle I_{6}=(\eta _{20}-\eta _{02})[(\eta _{30}+\eta _{12})^{2}-(\eta _{21}+\eta _{03})^{2}]+4\eta _{11}(\eta _{30}+\eta _{12})(\eta _{21}+\eta _{03})} I 7 = ( 3 η 21 − η 03 ) ( η 30 + η 12 ) [ ( η 30 + η 12 ) 2 − 3 ( η 21 + η 03 ) 2 ] − ( η 30 − 3 η 12 ) ( η 21 + η 03 ) [ 3 ( η 30 + η 12 ) 2 − ( η 21 + η 03 ) 2 ] . {\displaystyle I_{7}=(3\eta _{21}-\eta _{03})(\eta _{30}+\eta _{12})[(\eta _{30}+\eta _{12})^{2}-3(\eta _{21}+\eta _{03})^{2}]-(\eta _{30}-3\eta _{12})(\eta _{21}+\eta _{03})[3(\eta _{30}+\eta _{12})^{2}-(\eta _{21}+\eta _{03})^{2}].} These are well-known as Hu moment invariants. The first one, I1, is analogous to the moment of inertia around the image's centroid, where the pixels' intensities are analogous to physical density. The first six, I1 ... I6, are reflection symmetric, i.e. they are unchanged if the image is changed to a mirror image. The last one, I7, is reflection antisymmetric (changes sign under reflection), which enables it to distinguish mirror images of otherwise identical im
Artificial intelligence in marketing
Artificial intelligence marketing (AI marketing) is a form of marketing that uses artificial intelligence concepts and models such as machine learning, natural language processing, and computer vision to achieve marketing goals. The main difference between AI marketing and traditional forms of marketing reside in the reasoning, which is performed through a computer algorithm rather than a human. Each form of marketing has a different technique to the core of the marketing theory. Traditional marketing directly focuses on the needs of consumers; meanwhile some believe the shift AI may cause will lead marketing agencies to manage consumer needs instead. AI is used in various digital marketing spaces, such as content marketing, email marketing, online advertisement (in combination with machine learning), social media marketing, affiliate marketing, and beyond. == Historical development == AI in marketing has a long history, which goes all the way back to the 1980s. At this time, AI research was focusing on expert systems and robotics. Despite the initial research and the studies that were carried out, AI adoption remained limited. Research on it came to a stop for a while, until research was revived two decades later with the advancement in technology, the rise of big data, and a significant increase in computational power. Eventually, AI became very popular in the marketing world, and caught the eyes of many researchers as well as professionals. A large‐scale bibliometric study covering 1,580 peer‑reviewed papers published between 1982 and 2020 confirms that scholarly output on AI in marketing has surged since 2017, with Expert Systems with Applications emerging as the most prolific outlet. Prior to the application of artificial Intelligence in marketing, there was something called "collaborative filtering". This was used as early as 1998 by Amazon, and one of the first ways companies predicted consumer behavior, which enabled millions of recommendations to different customers. Personalized recommender systems are now widely used, for example to suggest music on Spotify, or TV shows on Netflix. A big milestone in AI marketing happened in 2014, when programmatic ad buying gained much greater popularity. Marketing consists of numerous manual tasks such as researching target markets, insertion orders, and managing high budgets as well as prices. In order to cut costs, and remove the need for these tedious tasks, many companies started to automate the marketing process with AI. In 2015, Google introduced RankBrain, a machine learning component of its search algorithm designed to interpret the intent behind user queries. RankBrain was followed by further AI-based search updates, including BERT in 2019, which improved the understanding of conversational queries, and the Multitask Unified Model (MUM) in 2021, which is multimodal and processes information across 75 languages. These advances shifted search engine optimization practice away from keyword matching toward content that satisfies user intent. Artificial intelligence is increasingly used in marketing to personalize user experiences and automate decision-making. For example, Netflix uses AI algorithms to recommend content based on viewing history, while Sephora employs chatbots to assist customers with product selection and availability. Programmatic advertising platforms like Google Ads leverage machine learning to optimize bidding strategies and target audiences more effectively. These applications demonstrate how AI enhances efficiency, engagement, and conversion rates across digital channels. === Artificial neural networks === An artificial neural network is a form of computer program modeled on the brain and nervous system of humans. Neural networks are composed of a series of interconnected processing neurons that function in unison to achieve certain outcomes. Using “human-like trial and error learning methods neural networks detect patterns existing within a data set ignoring data that is not significant while emphasizing the data which is most influential”. From a marketing perspective, neural networks are a form of software tool used to assist in decision making. Neural networks are effective in gathering and extracting information from large data sources and have the ability to identify cause and effect within tha data. These neural nets through the process of learning, identify relationships and connections between databases. Once knowledge has been accumulated, neural networks can be relied on to provide generalizations and can apply past knowledge and learning to a variety of situations. Neural networks help fulfill the role of marketing companies through effectively aiding in market segmentation and measurement of performance while reducing costs and improving accuracy. Due to their learning ability, flexibility, adaption, and knowledge discovery, neural networks offer many advantages over traditional models. Neural networks can be used to assist in pattern classification, forecasting and marketing analysis. == Tools and uses == Classification of customers can be facilitated through the neural network approach allowing companies to make informed marketing decisions. An example of this was employed by Spiegel Inc., a firm dealing in direct-mail operations that used neural networks to improve efficiencies. Using software developed by NeuralWare Inc., Spiegel identified the demographics of customers who had made a single purchase and those customers who had made repeat purchases. Neural networks where then able to identify the key patterns and consequently identify the customers that were most likely to repeat purchase. Understanding this information allowed Spiegel to streamline marketing efforts, and reduced costs. Sales forecasting “is the process of estimating future events with the goal of providing benchmarks for monitoring actual performance and reducing uncertainty". Artificial intelligence techniques have emerged to facilitate the process of forecasting through increasing accuracy in the areas of demand for products, distribution, employee turnover, performance measurement, and inventory control. An example of forecasting using neural networks is the Airline Marketing Assistant/Tactician; an application developed by BehabHeuristics which allows for the forecasting of passenger demand and consequent seat allocation through neural networks. This system has been used by National air Canada and USAir. Neural networks provide a useful alternative to traditional statistical models due to their reliability, time-saving characteristics and ability to recognize patterns from incomplete or noisy data. Examples of marketing analysis systems includes the Target Marketing System developed by Churchull Systems for Veratex Corporation. This support system scans a market database to identify dormant customers allowing management to make decisions regarding which key customers to target. When performing marketing analysis, neural networks can assist in the gathering and processing of information ranging from consumer demographics and credit history to the purchase patterns of consumers. Predictive analytics is a form of analytics involving the use of historical data and artificial intelligence algorithms to predict future trends and outcomes. It serves as a tool for anticipating and understanding user behavior based on patterns found in data. Predictive analytics uses artificial intelligence machine learning algorithms to recognize and predict patterns within data. Machine learning algorithms analyze the data, recognize patterns, and make predictions through continuous learning and adaptation. Predictive analytics is widely used across businesses and industries as a way to identify opportunities, avoid risks, and anticipate customer needs based on information derived from the analysis of user data. By analyzing historical customer data, artificial intelligence algorithms can deliver relevant and targeted marketing content. Recent systematic reviews show that generative large‑language models such as GPT‑3 and GPT‑4 are now routinely embedded in predictive‑analytics pipelines to mine unstructured market data and anticipate customer intent with greater precision. Personalization engines use artificial intelligence and machine learning to provide content or advertisements that are relevant to the user. User data is gathered, which then gets processed with machine learning, and patterns and trends among the users are identified. Users with shared characteristics or behaviors are then segmented into groups, and the personalization engine adjusts content and advertisements to match each segment's preferences. By processing a large amount of data, personalization engines are able to match users to advertisements and recommendations that align with their interests or preferences. Field evidence from consumer‑goods and electronics firms indicates that AI‑driven personalization can raise