AI Writing Tools

Explore the best AI Writing Tools — independent reviews, comparisons, pricing and step-by-step how-to guides, curated by Aizhi.

• 

  Adrozek

  Adrozek is malware that injects fake ads into online search results. Microsoft announced the malware threat on 10 December 2020, and noted that many different browsers are affected, including Google Chrome, Microsoft Edge, Mozilla Firefox and Yandex Browser. The malware was first detected in May 2020 and, at its peak in August 2020, controlled over 30,000 devices a day. But during the December 2020 announcement, Microsoft claimed "hundreds of thousands" of infected devices worldwide between May and September 2020. According to Microsoft, if not detected and blocked, Adrozek adds browser extensions, modifies a specific DLL per target browser, and changes browser settings to insert additional, unauthorized ads into web pages, often on top of legitimate ads from search engines. For each user tricked into clicking on the fake ads, the scammers earn affiliate advertising dollars. The malware has been observed to extract device data and, in some cases, steal credentials, sending them to remote servers. Users may unintentionally install the malware because of a drive-by download, by visiting a tampered website, opening an e-mail attachment, or clicking on a deceptive link or a deceptive pop-up window. The main malware program is downloaded to the “Programs Files” folder using file names such as Audiolava.exe, QuickAudio.exe, and converter.exe. According to PC Magazine, a good way to avoid, or mitigate, infection by Adrozek is to keep browser and related software programs up to date.

  Read more →
• 

  Learning vector quantization

  In computer science, learning vector quantization (LVQ) is a prototype-based supervised classification algorithm. LVQ is the supervised counterpart of vector quantization systems. LVQ can be understood as a special case of an artificial neural network, more precisely, it applies a winner-take-all Hebbian learning-based approach. It is a precursor to self-organizing maps (SOM) and related to neural gas and the k-nearest neighbor algorithm (k-NN). LVQ was invented by Teuvo Kohonen. == Definition == An LVQ system is represented by prototypes W = ( w ( i ) , . . . , w ( n ) ) {\displaystyle W=(w(i),...,w(n))} which are defined in the feature space of observed data. In winner-take-all training algorithms one determines, for each data point, the prototype which is closest to the input according to a given distance measure. The position of this so-called winner prototype is then adapted, i.e. the winner is moved closer if it correctly classifies the data point or moved away if it classifies the data point incorrectly. An advantage of LVQ is that it creates prototypes that are easy to interpret for experts in the respective application domain. LVQ systems can be applied to multi-class classification problems in a natural way. A key issue in LVQ is the choice of an appropriate measure of distance or similarity for training and classification. Recently, techniques have been developed which adapt a parameterized distance measure in the course of training the system, see e.g. (Schneider, Biehl, and Hammer, 2009) and references therein. LVQ can be a valuable aid in classifying text documents. == Algorithm == The algorithms are presented as in. Set up: Let the data be denoted by x i ∈ R D {\displaystyle x_{i}\in \mathbb {R} ^{D}} , and their corresponding labels by y i ∈ { 1 , 2 , … , C } {\displaystyle y_{i}\in \{1,2,\dots ,C\}} . The complete dataset is { ( x i , y i ) } i = 1 N {\displaystyle \{(x_{i},y_{i})\}_{i=1}^{N}} . The set of code vectors is w j ∈ R D {\displaystyle w_{j}\in \mathbb {R} ^{D}} . The learning rate at iteration step t {\displaystyle t} is denoted by α t {\displaystyle \alpha _{t}} . The hyperparameters w {\displaystyle w} and ϵ {\displaystyle \epsilon } are used by LVQ2 and LVQ3. The original paper suggests ϵ ∈ [ 0.1 , 0.5 ] {\displaystyle \epsilon \in [0.1,0.5]} and w ∈ [ 0.2 , 0.3 ] {\displaystyle w\in [0.2,0.3]} . === LVQ1 === Initialize several code vectors per label. Iterate until convergence criteria is reached. Sample a datum x i {\displaystyle x_{i}} , and find out the code vector w j {\displaystyle w_{j}} , such that x i {\displaystyle x_{i}} falls within the Voronoi cell of w j {\displaystyle w_{j}} . If its label y i {\displaystyle y_{i}} is the same as that of w j {\displaystyle w_{j}} , then w j ← w j + α t ( x i − w j ) {\displaystyle w_{j}\leftarrow w_{j}+\alpha _{t}(x_{i}-w_{j})} , otherwise, w j ← w j − α t ( x i − w j ) {\displaystyle w_{j}\leftarrow w_{j}-\alpha _{t}(x_{i}-w_{j})} . === LVQ2 === LVQ2 is the same as LVQ3, but with this sentence removed: "If w j {\displaystyle w_{j}} and w k {\displaystyle w_{k}} and x i {\displaystyle x_{i}} have the same class, then w j ← w j − α t ( x i − w j ) {\displaystyle w_{j}\leftarrow w_{j}-\alpha _{t}(x_{i}-w_{j})} and w k ← w k + α t ( x i − w k ) {\displaystyle w_{k}\leftarrow w_{k}+\alpha _{t}(x_{i}-w_{k})} .". If w j {\displaystyle w_{j}} and w k {\displaystyle w_{k}} and x i {\displaystyle x_{i}} have the same class, then nothing happens. === LVQ3 === Initialize several code vectors per label. Iterate until convergence criteria is reached. Sample a datum x i {\displaystyle x_{i}} , and find out two code vectors w j , w k {\displaystyle w_{j},w_{k}} closest to it. Let d j := ‖ x i − w j ‖ , d k := ‖ x i − w k ‖ {\displaystyle d_{j}:=\|x_{i}-w_{j}\|,d_{k}:=\|x_{i}-w_{k}\|} . If min ( d j d k , d k d j ) > s {\displaystyle \min \left({\frac {d_{j}}{d_{k}}},{\frac {d_{k}}{d_{j}}}\right)>s} , where s = 1 − w 1 + w {\displaystyle s={\frac {1-w}{1+w}}} , then If w j {\displaystyle w_{j}} and x i {\displaystyle x_{i}} have the same class, and w k {\displaystyle w_{k}} and x i {\displaystyle x_{i}} have different classes, then w j ← w j + α t ( x i − w j ) {\displaystyle w_{j}\leftarrow w_{j}+\alpha _{t}(x_{i}-w_{j})} and w k ← w k − α t ( x i − w k ) {\displaystyle w_{k}\leftarrow w_{k}-\alpha _{t}(x_{i}-w_{k})} . If w k {\displaystyle w_{k}} and x i {\displaystyle x_{i}} have the same class, and w j {\displaystyle w_{j}} and x i {\displaystyle x_{i}} have different classes, then w j ← w j − α t ( x i − w j ) {\displaystyle w_{j}\leftarrow w_{j}-\alpha _{t}(x_{i}-w_{j})} and w k ← w k + α t ( x i − w k ) {\displaystyle w_{k}\leftarrow w_{k}+\alpha _{t}(x_{i}-w_{k})} . If w j {\displaystyle w_{j}} and w k {\displaystyle w_{k}} and x i {\displaystyle x_{i}} have the same class, then w j ← w j − ϵ α t ( x i − w j ) {\displaystyle w_{j}\leftarrow w_{j}-\epsilon \alpha _{t}(x_{i}-w_{j})} and w k ← w k + ϵ α t ( x i − w k ) {\displaystyle w_{k}\leftarrow w_{k}+\epsilon \alpha _{t}(x_{i}-w_{k})} . If w k {\displaystyle w_{k}} and x i {\displaystyle x_{i}} have different classes, and w j {\displaystyle w_{j}} and x i {\displaystyle x_{i}} have different classes, then the original paper simply does not explain what happens in this case, but presumably nothing happens in this case. Otherwise, skip. Note that condition min ( d j d k , d k d j ) > s {\displaystyle \min \left({\frac {d_{j}}{d_{k}}},{\frac {d_{k}}{d_{j}}}\right)>s} , where s = 1 − w 1 + w {\displaystyle s={\frac {1-w}{1+w}}} , precisely means that the point x i {\displaystyle x_{i}} falls between two Apollonian spheres.

  Read more →
• 

  John M. Jumper

  John Michael Jumper (born 1 January 1985) is an American chemist and computer scientist. Jumper and Demis Hassabis were awarded the 2024 Nobel Prize in Chemistry for protein structure prediction. As of 2025 Jumper serves as director at Google DeepMind. Jumper and his colleagues created AlphaFold, an artificial intelligence (AI) model to predict protein structures from their amino acid sequence with high accuracy. The AlphaFold team had released 214 million protein structures as of January 2024. The scientific journal Nature included Jumper as one of the ten "people who mattered" in science in their annual listing of Nature's 10 in 2021. == Education == Jumper graduated from Pulaski Academy in 2003. He received a Bachelor of Science with majors in physics and mathematics from Vanderbilt University in 2007, a Master of Philosophy in theoretical condensed matter physics from the University of Cambridge where he was a student of St Edmund's College, Cambridge in 2010 on a Marshall Scholarship, a Master of Science in theoretical chemistry from the University of Chicago in 2012, and a Doctor of Philosophy in theoretical chemistry from the University of Chicago in 2017. His doctoral advisors at the University of Chicago were Tobin R. Sosnick and Karl Freed. == Career and research == Jumper's research investigates algorithms for protein structure prediction. === AlphaFold === AlphaFold is a deep learning algorithm developed by Jumper and his team at DeepMind, a research lab acquired by Google's parent company Alphabet Inc. It is an artificial intelligence program which performs predictions of protein structure. === Awards and honors === In November 2020, AlphaFold was named the winner of the 14th Critical Assessment of Structure Prediction (CASP) competition. This international competition benchmarks algorithms to determine which one can best predict the 3D structure of proteins. AlphaFold won the competition, outperforming other algorithms scoring above 90 for around two-thirds of the proteins in CASP's global distance test (GDT), a test that measures the degree to which a computational program predicted structure is similar to the lab experiment determined structure, with 100 being a complete match, within the distance cutoff used for calculating GDT. In 2021, Jumper was awarded the BBVA Foundation Frontiers of Knowledge Award in the category "Biology and Biomedicine". In 2022 Jumper received the Wiley Prize in Biomedical Sciences and for 2023 the Breakthrough Prize in Life Sciences for developing AlphaFold, which accurately predicts the structure of a protein. In 2023 he was awarded the Canada Gairdner International Award and the Albert Lasker Award for Basic Medical Research. In 2024, Jumper and Demis Hassabis shared half of the Nobel Prize in Chemistry for their protein folding predictions, the other half went to David Baker for computational protein design. In 2025, Jumper received the Golden Plate Award of the American Academy of Achievement and the Marshall Medal of the Marshall Aid Commemoration Commission. He was elected a Fellow of the Royal Society (FRS) that same year. In 2026, he was elected a member of the National Academy of Engineering.

  Read more →
• 

  Mittens (chess)

  Mittens is a chess engine developed by Chess.com. It was released on January 1, 2023, alongside four other engines, all of them given cat-related names. The engine became a viral sensation in the chess community due to exposure through content made by chess streamers and a social media marketing campaign, later contributing to record levels of traffic to the Chess.com website and causing issues with database scalability. Mittens was given a rating of one point by Chess.com, although it was evidently stronger than that. Various chess masters played matches against the engine, with players such as Hikaru Nakamura and Levy Rozman drawing and losing their games respectively. A month after its release, Mittens was removed from the website on February 1, as expected through Chess.com's monthly bot cycles. In December 2023, Mittens was brought back in a group of Chess.com's most popular bots of 2023. In January 2024, Mittens was removed again. == Release == Mittens was released on January 1, 2023, as part of a New Year event on Chess.com. It was one of five engines released, all with names related to cats. The other engines released were named Scaredy Cat, rated 800; Angry Cat, rated 1000; Mr. Grumpers, rated 1200 and Catspurrov (a pun on Garry Kasparov), rated 1400. As part of the announcement, a picture of each engine was accompanied by a short description of its character. The description given for Mittens suggested that the engine was hiding something, reading: Mittens likes chess… But how good is she? Of the five engines released, Mittens was by far the most popular. In December 2023, Chess.com re-released Mittens as part of a "best of 2023" group of chess bots made to showcase their most popular bots of the year. == Design == Mittens was conceptualized by Chess.com employee Will Whalen. Appearing as a kitten, Mittens trash talked its opponents with a selection of voice lines: these lines included quotes from J. Robert Oppenheimer, Vincent van Gogh and Friedrich Nietzsche, as well as the 1967 film Le Samouraï. The engine's "personality" was devised by a writing team headed by Sean Becker, and Marija Casic provided the engine's graphics. Chess.com did not disclose any information about the software running the engine. It may be based on Chess.com's Komodo Dragon 3 engine. Mittens' strategy was to slowly grind down an opponent, a tactic likened to the playing style of Anatoly Karpov. Becker stated that the design team believed it would be "way more demoralizing and funny" for the engine to play this way. According to Hikaru Nakamura, Mittens sometimes missed the best move (or winning positions). == Rating == On Chess.com, Mittens had a rating of one point. However, the engine's playing style and tactics showed that it was stronger than that; Mittens was able to beat or draw against many top human players. In an interview with CNN Business, Whalen stated that the idea behind giving Mittens a rating of one was to surprise its opponents, giving it the upper hand psychologically. Estimates of Mittens' true rating range from an Elo of 3200 to 3500, because of its ability to beat other engines of around that level. An upper bound of the engine's rating was found after Levy Rozman made Mittens play against Stockfish 15, a 3700 rated engine. Mittens lost the two games that the engines played. The range of Mittens' possible ratings was summarized by Dot Esports, who stated: It seems like she’s around the 3200–3500 rating range (in Chess.com terms, where the best human players, like Magnus Carlsen and Hikaru Nakamura, sport a 3000–3100 rating in the faster formats), as evidenced by her victories over the site’s otherwise strongest, 3200-rated bots, and her defeat to Stockfish 15, which is currently rated around 3700. == Games == Against human players, Mittens won over 99 percent of the millions of games it played. Chess players such as Hikaru Nakamura, Benjamin Bok, Levy Rozman and Eric Rosen struggled against Mittens; while Rozman and Rosen both lost against the engine, Nakamura and Bok were both able to make a draw. In particular, Nakamura's game against the engine lasted 166 moves; he was playing as White. Bok, Benjamin Finegold and Rozman later went on to win against Mittens, the latter with engine assistance from Stockfish. Magnus Carlsen publicly refused to play the engine, calling it a "transparent marketing trick" and "a soulless computer". Against other chess engines, Mittens participated in the Chess.com Computer Chess Championship as a side act. In the competition, Mittens played 150 games against an engine named after the film M3GAN and won overall with a score of 81.5 to 68.5. This equated to 54 percent of the games played. During the event, an estimate of Mittens' rating was made at 3515 points. == Impact == Mittens went viral in the chess community due to its concept and design: according to an announcement by Chess.com, a combined total of 120 million games were played against the cat engines over the course of January, with around 40 million played against Mittens. The popularity of the engine was helped by the social media exposure created by Chess.com. This included creating an official Twitter account to promote the engine. Chess streamers like Rozman and Nakamura helped cultivate this by creating content around the engine. A video by Nakamura entitled "Mittens the chess bot will make you quit chess" gained over 3.5 million views on YouTube. On January 11, Chess.com reported issues with database scalability due to record levels of traffic: 40 percent more games had been played on Chess.com in January 2023 than any other month since the website's release. According to The Wall Street Journal, the popularity spike was more than the similar surge following the release of Netflix's The Queen's Gambit. The popularity of Mittens was cited by Chess.com as a reason for this instability. The problems continued throughout January; Chess.com stated that they would have to upgrade their servers and invest more in cloud computing to solve the problems caused by the website's popularity surge. On February 1, 2023, Mittens and the other cat engines were removed from the computer section of Chess.com. They were replaced with five new engines themed around artificial intelligence. A tweet was posted on the Mittens's Twitter account after the engine's removal, reading "This is just the beginning. Goodbye for now."

  Read more →
• 

  Site Security Handbook

  The Site Security Handbook, RFC 2196, is a guide on setting computer security policies and procedures for sites that have systems on the Internet (however, the information provided should also be useful to sites not yet connected to the Internet). The guide lists issues and factors that a site must consider when setting their own policies. It makes a number of recommendations and provides discussions of relevant areas. This guide is only a framework for setting security policies and procedures. In order to have an effective set of policies and procedures, a site will have to make many decisions, gain agreement, and then communicate and implement these policies. The guide is a product of the IETF SSH working group, and was published in 1997, obsoleting the earlier RFC 1244 from 1991.

  Read more →
• 

  Computer game bot Turing test

  The computer game bot Turing test is a variant of the Turing test, where a human judge viewing and interacting with a virtual world must distinguish between other humans and video game bots, both interacting with the same virtual world. This variant was first proposed in 2008 by Associate Professor Philip Hingston of Edith Cowan University, and implemented through a tournament called the 2K BotPrize. == History == The computer game bot Turing test was proposed to advance the fields of artificial intelligence (AI) and computational intelligence with respect to video games. It was considered that a poorly implemented bot implied a subpar game, so a bot that would be capable of passing this test, and therefore might be indistinguishable from a human player, would directly improve the quality of a game. It also served to debunk a flawed notion that "game AI is a solved problem." Emphasis is placed on a game bot that interacts with other players in a multiplayer environment. Unlike a bot that simply needs to make optimal human-like decisions to play or beat a game, this bot must make the same decisions while also convincing another in-game player of its human-likeness. == Implementation == The computer game bot Turing test was designed to test a bot's ability to interact with a game environment in comparison with a human player; simply 'winning' was insufficient. This evolved into a contest with a few important goals in mind: There are three participants: a human player, a computer-game bot, and a judge. The bot needs to appear more human-like than the human player. Judge scores are not bipolar — both human and bot can be scored anywhere on a scale from 1 to 5 (1=not humanlike, 5=human). All three participants are to be indistinguishable in the arena, with the exception of a randomly generated name tag, so as to reduce the chance of random elements such as name or appearance influencing the judges. Chat is disabled throughout the match. Bots were not given omniscient powers as they may be in other games. Bots must react only to the data that might be reasonably available to a human player. Human participants were of a moderate skill range, with no participant either ignorant to the game or capable of playing at a professional level. In 2008, the first 2K BotPrize tournament took place. The contest was held with the game Unreal Tournament 2004 as the platform. Contestants created their bots in advance using the GameBots interface. GameBots had some modifications made so as to adhere to the above conditions, such as removing data about vantage points or weapon damage that unfairly informed the bots of relevant strengths/weaknesses that a human would otherwise need to learn. == Tournament == The first BotPrize Tournament was held on 17 December 2008, as part of the 2008 IEEE Symposium on Computational Intelligence and Games in Australia. Each competing team was given time to set up and adjust their bots to the modified game client, although no coding changes were allowed at that point. The tournament was run in rounds, each a 10-minute death match. Judges were the last to join the server and every judge observed every player and every bot exactly once, although the pairing of players and bots did change. When the tournament ended, no bot was rated as more human than any player. In subsequent tournaments, run during 2009–2011, bots achieved scores that were increasingly human-like, but no contestant had won the BotPrize in any of these contests. In 2012, the 2K BotPrize was held once again, and two teams programmed bots that achieved scores greater than those of human players. == Successful bots == To date, there have been two successfully programmed bots that passed the computer game bot Turing test: UT^2, a team from the University of Texas at Austin, emphasized a bot that adjusted its behaviour based on previously observed human behaviour and neuroevolution. The team has made their bot available, although a copy of Unreal Tournament 2004 is required. Mihai Polceanu, a doctoral student from Romania, focused on creating a bot that would mimic opponent reactions, in a sense 'borrowing' the human-like nature of the opponent. These victors succeeded in the year 2012, Alan Turing's centenary year. == Aftermath == The outcome of a bot that appears more human-like than a human player is possibly overstated, since in the tournament in which the bots succeeded, the average 'humanness' rating of the human players was only 41.4%. This showcases some limits of this Turing test, since the results demonstrate that human behaviour is more complicated and quantitative than was accounted for. In light of this, the BotPrize competition organizers will increase the difficulty in upcoming years with new challenges, forcing competitors to improve their bots. It is also believed that methods and techniques developed for the computer game bot Turing test will be useful in fields other than video games, such as virtual training environments and in improving Human–robot interaction. == Contrasts to the Turing test == The computer game bot Turing test differs from the traditional or generic Turing test in a number of ways: Unlike the traditional Turing test, for example the Chatterbot-style contest held annually by the Loebner Prize competition, the humans who played against the Computer Game Bots are not trying to convince judges they are the human; rather, they want to win the game (i.e., by achieving the highest kill score). Judges are not restricted to awarding only one participant in a match as the 'human' and the other as the 'non-human.' This emphasizes more qualitative rather than polarized findings. With regards to a successful video game bot, this is not to be confused with a claim that the bot is 'intelligent,' whereas a machine that 'passed' the Turing test would arguably have some evidence for its Chatterbot's 'intelligence.' The game Unreal Tournament 2004 was chosen for its commercial availability and its interface for creating bots, GameBots. This limitation on medium is a sharp contrast to the Turing test, which emphasizes a conversation, where possible questions are vastly more numerous than the set of possible actions available in any specific video game. The available information to the participants, humans and bots, is not equal. Humans interact through vision and sound, whereas bots interact with data and events. The judges cannot introduce new events (e.g., a lava pit) to aid in differentiating between human and bot, whereas in a Chatterbot designed system, judges may theoretically ask any question in any manner. The two participants and the judge take part in a three-way interaction, unlike, for example, the paired two-way interaction of the Loebner Prize Contest.

  Read more →
• 

  Mathematical model

  A mathematical model is an abstract description of a concrete system using mathematical concepts and language. The process of developing a mathematical model is termed mathematical modeling. Mathematical models are used in many fields, including applied mathematics, natural sciences, social sciences and engineering. In particular, the field of operations research studies the use of mathematical modelling and related tools to solve problems in business or military operations. A model may help to characterize a system by studying the effects of different components, which may be used to make predictions about behavior or solve specific problems. == Elements of a mathematical model == Mathematical models can take many forms, including dynamical systems, statistical models, differential equations, or game theoretic models. These and other types of models can overlap, with a given model involving a variety of abstract structures. In many cases, the quality of a scientific field depends on how well the mathematical models developed on the theoretical side agree with results of repeatable experiments. Lack of agreement between theoretical mathematical models and experimental measurements often leads to important advances as better theories are developed. In the physical sciences, a traditional mathematical model contains most of the following elements: Governing equations Supplementary sub-models Defining equations Constitutive equations Assumptions and constraints Initial and boundary conditions Classical constraints and kinematic equations == Classifications == Mathematical models are of different types: === Linear vs. nonlinear === If all the operators in a mathematical model exhibit linearity, the resulting mathematical model is defined as linear. All other models are considered nonlinear. The definition of linearity and nonlinearity is dependent on context, and linear models may have nonlinear expressions in them. For example, in a statistical linear model, it is assumed that a relationship is linear in the parameters, but it may be nonlinear in the predictor variables. Similarly, a differential equation is said to be linear if it can be written with linear differential operators, but it can still have nonlinear expressions in it. In a mathematical programming model, if the objective functions and constraints are represented entirely by linear equations, then the model is regarded as a linear model. If one or more of the objective functions or constraints are represented with a nonlinear equation, then the model is known as a nonlinear model. Linear structure implies that a problem can be decomposed into simpler parts that can be treated independently or analyzed at a different scale, and therefore that the results will remain valid if the initial is recomposed or rescaled. Nonlinearity, even in fairly simple systems, is often associated with phenomena such as chaos and irreversibility. Although there are exceptions, nonlinear systems and models tend to be more difficult to study than linear ones. A common approach to nonlinear problems is linearization, but this can be problematic if one is trying to study aspects such as irreversibility, which are strongly tied to nonlinearity. === Static vs. dynamic === A dynamic model accounts for time-dependent changes in the state of the system, while a static (or steady-state) model calculates the system in equilibrium, and thus is time-invariant. Dynamic models are typically represented by differential equations or difference equations. === Explicit vs. implicit === If all of the input parameters of the overall model are known, and the output parameters can be calculated by a finite series of computations, the model is said to be explicit. But sometimes it is the output parameters which are known, and the corresponding inputs must be solved for by an iterative procedure, such as Newton's method or Broyden's method. In such a case the model is said to be implicit. For example, a jet engine's physical properties such as turbine and nozzle throat areas can be explicitly calculated given a design thermodynamic cycle (air and fuel flow rates, pressures, and temperatures) at a specific flight condition and power setting, but the engine's operating cycles at other flight conditions and power settings cannot be explicitly calculated from the constant physical properties. === Discrete vs. continuous === A discrete model treats objects as discrete, such as the particles in a molecular model or the states in a statistical model; while a continuous model represents the objects in a continuous manner, such as the velocity field of fluid in pipe flows, temperatures and stresses in a solid, and electric field that applies continuously over the entire model due to a point charge. === Deterministic vs. probabilistic (stochastic) === A deterministic model is one in which every set of variable states is uniquely determined by parameters in the model and by sets of previous states of these variables; therefore, a deterministic model always performs the same way for a given set of initial conditions. Conversely, in a stochastic model—usually called a "statistical model"—randomness is present, and variable states are not described by unique values, but rather by probability distributions. === Deductive, inductive, or floating === A deductive model is a logical structure based on a theory. An inductive model arises from empirical findings and generalization from them. If a model rests on neither theory nor observation, it may be described as a 'floating' model. Application of mathematics in social sciences outside of economics has been criticized for unfounded models. Application of catastrophe theory in science has been characterized as a floating model. === Strategic vs. non-strategic === Models used in game theory are distinct in the sense that they model agents with incompatible incentives, such as competing species or bidders in an auction. Strategic models assume that players are autonomous decision makers who rationally choose actions that maximize their objective function. A key challenge of using strategic models is defining and computing solution concepts such as the Nash equilibrium. An interesting property of strategic models is that they separate reasoning about rules of the game from reasoning about behavior of the players. == Construction == In business and engineering, mathematical models may be used to maximize a certain output. The system under consideration will require certain inputs. The system relating inputs to outputs depends on other variables too: decision variables, state variables, exogenous variables, and random variables. Decision variables are sometimes known as independent variables. Exogenous variables are sometimes known as parameters or constants. The variables are not independent of each other as the state variables are dependent on the decision, input, random, and exogenous variables. Furthermore, the output variables are dependent on the state of the system (represented by the state variables). Objectives and constraints of the system and its users can be represented as functions of the output variables or state variables. The objective functions will depend on the perspective of the model's user. Depending on the context, an objective function is also known as an index of performance, as it is some measure of interest to the user. Although there is no limit to the number of objective functions and constraints a model can have, using or optimizing the model becomes more involved (computationally) as the number increases. For example, economists often apply linear algebra when using input–output models. Complicated mathematical models that have many variables may be consolidated by use of vectors where one symbol represents several variables. === A priori information === Mathematical modeling problems are often classified into black box or white box models, according to how much a priori information on the system is available. A black-box model is a system of which there is no a priori information available. A white-box model (also called glass box or clear box) is a system where all necessary information is available. Practically all systems are somewhere between the black-box and white-box models, so this concept is useful only as an intuitive guide for deciding which approach to take. Usually, it is preferable to use as much a priori information as possible to make the model more accurate. Therefore, the white-box models are usually considered easier, because if you have used the information correctly, then the model will behave correctly. Often the a priori information comes in forms of knowing the type of functions relating different variables. For example, if we make a model of how a medicine works in a human system, we know that usually the amount of medicine in the blood is an exponentially decaying function, but we are still left with several unknown parameters; how

  Read more →
• 

  Cerebellar model articulation controller

  The cerebellar model arithmetic computer (CMAC) is a type of neural network based on a model of the mammalian cerebellum. It is also known as the cerebellar model articulation controller. It is a type of associative memory. The CMAC was first proposed as a function modeler for robotic controllers by James Albus in 1975 (hence the name), but has been extensively used in reinforcement learning and also as for automated classification in the machine learning community. The CMAC is an extension of the perceptron model. It computes a function for n {\displaystyle n} input dimensions. The input space is divided up into hyper-rectangles, each of which is associated with a memory cell. The contents of the memory cells are the weights, which are adjusted during training. Usually, more than one quantisation of input space is used, so that any point in input space is associated with a number of hyper-rectangles, and therefore with a number of memory cells. The output of a CMAC is the algebraic sum of the weights in all the memory cells activated by the input point. A change of value of the input point results in a change in the set of activated hyper-rectangles, and therefore a change in the set of memory cells participating in the CMAC output. The CMAC output is therefore stored in a distributed fashion, such that the output corresponding to any point in input space is derived from the value stored in a number of memory cells (hence the name associative memory). This provides generalisation. == Building blocks == In the adjacent image, there are two inputs to the CMAC, represented as a 2D space. Two quantising functions have been used to divide this space with two overlapping grids (one shown in heavier lines). A single input is shown near the middle, and this has activated two memory cells, corresponding to the shaded area. If another point occurs close to the one shown, it will share some of the same memory cells, providing generalisation. The CMAC is trained by presenting pairs of input points and output values, and adjusting the weights in the activated cells by a proportion of the error observed at the output. This simple training algorithm has a proof of convergence. It is normal to add a kernel function to the hyper-rectangle, so that points falling towards the edge of a hyper-rectangle have a smaller activation than those falling near the centre. One of the major problems cited in practical use of CMAC is the memory size required, which is directly related to the number of cells used. This is usually ameliorated by using a hash function, and only providing memory storage for the actual cells that are activated by inputs. == One-step convergent algorithm == Initially least mean square (LMS) method is employed to update the weights of CMAC. The convergence of using LMS for training CMAC is sensitive to the learning rate and could lead to divergence. In 2004, a recursive least squares (RLS) algorithm was introduced to train CMAC online. It does not need to tune a learning rate. Its convergence has been proved theoretically and can be guaranteed to converge in one step. The computational complexity of this RLS algorithm is O(N3). == Hardware implementation infrastructure == Based on QR decomposition, an algorithm (QRLS) has been further simplified to have an O(N) complexity. Consequently, this reduces memory usage and time cost significantly. A parallel pipeline array structure on implementing this algorithm has been introduced. Overall by utilizing QRLS algorithm, the CMAC neural network convergence can be guaranteed, and the weights of the nodes can be updated using one step of training. Its parallel pipeline array structure offers its great potential to be implemented in hardware for large-scale industry usage. == Continuous CMAC == Since the rectangular shape of CMAC receptive field functions produce discontinuous staircase function approximation, by integrating CMAC with B-splines functions, continuous CMAC offers the capability of obtaining any order of derivatives of the approximate functions. == Deep CMAC == In recent years, numerous studies have confirmed that by stacking several shallow structures into a single deep structure, the overall system could achieve better data representation, and, thus, more effectively deal with nonlinear and high complexity tasks. In 2018, a deep CMAC (DCMAC) framework was proposed and a backpropagation algorithm was derived to estimate the DCMAC parameters. Experimental results of an adaptive noise cancellation task showed that the proposed DCMAC can achieve better noise cancellation performance when compared with that from the conventional single-layer CMAC. == Summary ==

  Read more →
• 

  Stride (software)

  Stride was a cloud-based team business communication and collaboration tool, launched by Atlassian on 7 September 2017 to replace the cloud-based version of HipChat. Stride software was available to download onto computers running Windows, Mac or Linux, as well as Android, iOS smartphones, and tablets. Stride was bought by Atlassian's competitor Slack Technologies and was discontinued on February 15, 2019. The features of Stride include chat rooms, one-on-one messaging, file sharing, 5 GB of file storage, group voice and video calling, built-in collaboration tools, and up to 25,000 of searchable message history. Premium features include unlimited file storage, users, group chat rooms, file sharing and storage, apps, and history retention. The premium version, priced at $3/user/month, also includes advanced meeting functionality like group screen sharing, remote desktop control, and dial-in/dial-out capabilities. Stride offered integrations with Atlassian's other products as well as other third-party applications listed in the Atlassian Marketplace, such as GitHub, Giphy, Stand-Bot and Google Calendar. Stride offered additional features beyond messaging to improve efficiency and productivity. It aimed to reduce collaboration noise by introducing a "focus" mode, and eliminates the divisions between text chat, voice meetings, and videoconferencing, by simplifying transitioning between these modes in the same channel. On July 26, 2018, Atlassian announced that HipChat and Stride would be discontinued February 15, 2019, and that it had reached a deal to sell their intellectual property to Slack. Slack paid an undisclosed amount over three years to assume the user bases of the services, while Atlassian took a minority investment in Slack. The companies also announced a commitment to work on integration of Slack with Atlassian services.

  Read more →
• 

  National Security Memorandum on Artificial Intelligence

  The Memorandum on Advancing the United States' Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificial Intelligence is a memorandum signed by U.S. president Joe Biden. The memorandum is described as seeking to advance U.S. leadership in the development of safe, secure, and trustworthy artificial intelligence (AI); enable the U.S. government to use AI for national security; and contribute to international AI governance.

  Read more →
• 

  Korean Decimal Classification

  The Korean Decimal Classification (KDC) is a system of library classification used in South Korea. The structure and main level classes of the KDC are based on the Dewey Decimal Classification. The KDC is maintained and published by the Classification Committee of the Korean Library Association. The first edition of the classification was published in 1964; the most recent edition is the sixth edition published in 2013. Almost all school and public libraries in South Korea use the KDC to organize their collections, as well as the National Library of Korea and some university libraries. == History == Multiple library classification systems had been developed for Korean libraries before the publication of the KDC. These included the Railway Bureau Library Classification(1920), the Korean Decimal Classification edited by Bong-Suk Park(known as KDCP, 1947), the Han-Un Decimal Classification(1954), and the Kuk-Yeon Decimal Classification(1958). After the disappearance of editor Bong-Suk Park in the 1950s, the KDCP system decreased in use. Korean librarians considered adopting the Dewey Decimal Classification (DDC), especially after it was implemented at Yonsei University in 1957, but struggled to apply it to East Asian and Korean-focused works in their collections. In February 1963, members of the Korean Library Association's Classification were appointed to create a national classification; they decided to make revisions to the order of the main classes of the DDC, for example bringing together the class Language(700) together with the class for Literature(800). Committee members prepared draft classes and indexes and the first edition of the KDC was published in May 1964. Both the text and the index were written in Korean Hangul characters and Chinese characters. The second edition was published just two years later, in 1966, correcting errors and omissions found in the first edition. The third edition was published in 1980, maintaining the basic framework of the previous editions while expanding significantly. The fourth edition, published in 1996, made considerable changes, including increasing the number of representatives on the Classification Committee. The committee sought feedback from the library community and implemented revisions included in the recently published edition 20 of the DDC and edition 9 of the Nippon Decimal Classification. New policies applied to the fourth edition included principles suggesting the main classes should remain as static as possible, with focus shown to expanding classes devoted to technology and science. Likewise, many subject specialists were consulted for the publication of the fifth edition in 2009. The publication of the 23rd edition of the DDC in 2011 provided opportunity for a new revision of the KDC, and the sixth edition was published in July 2013. Greater numbers of classes provided number building capacity in the sixth edition, allowing for more specificity. == Description == The KDC classifies resources primarily by discipline, though some classes are collocated by subject. There are eight auxiliary mnemonic tables used to expand class numbers. The main classes of the KDC are the same as the main classes of the Dewey Decimal Classification, but four of those main classes are in a different order: Natural sciences (400), Technology and engineering (500), Arts (600), and Language 700. Though the structure is heavily influenced by the DDC, aspects of multiple library classifications have been invoked in the creation of the KDC, including the Library of Congress Classification for the arrangement of the social sciences (300), the Universal Decimal Classification for medical sciences (510), the KDCP for Korean and Oriental subjects, the Nippon Decimal Classification for those of Japan and Oriental subjects. === Classes of the KDC 6th edition === 000 General works 000 General works 010 Books, Bibliography 020 Library & information science 030 General encyclopedias 040 General collected essays 050 General serial publications 060 General societies 070 Newspapers, journalism 080 General collected works 090 Materials of province 100 Philosophy 100 Philosophy 110 Metaphysics 120 Epistemology, etc. 130 Systems of philosophy 140 Chinese classics 150 Oriental philosophy and thought 160 Western philosophy 170 Logic 180 Psychology 190 Ethics, moral philosophy 200 Religion 200 Religion 210 Comparative religion 220 Buddhism 230 Christian religion 240 Taoism 250 Chondoism 260 [Unassigned] 270 Hinduism, Brahmanism 280 Islam, Mohammedianism 290 Other religions 300 Social sciences 300 Social sciences 310 Statistics 320 Economics 330 Sociology and social problems 340 Political sciences 350 Public administration 360 Law 370 Education 380 Customs, Etiquette, Folklore 390 Military science 400 Natural sciences 400 Natural sciences 410 Mathematics 420 Physics 430 Chemistry 440 Astronomy 450 Earth science 460 Mineralogy 470 Life science 480 Botany 490 Zoological science 500 Technology 500 Technology 510 Medical science 520 Agriculture 530 Engineering, technology, etc. 540 Construction and architecture 550 Mechanical engineering 560 Electrical, comm. & electric engineering 570 Chemical engineering 580 Manufactures 590 Human ecology 600 Arts 600 Arts 610 [Unassigned] 620 Sculpture, plastic art 630 Crafts 640 Calligraphy 650 Painting, design 660 Photography 670 Music 680 Stage performance, museum arts 690 Amusements, sports & physical training 700 Language 700 Language 710 Korean language 720 Chinese language 730 Japanese & other Asian languages 740 English 750 German 760 French languages 770 Spanish languages & Portuguese language 780 Italian languages 790 Other languages 800 Literature 800 Literature 810 Korean literature 820 Chinese literature 830 Japanese & other Asian literature 840 English & American literature 850 German literature 860 French literature 870 Spanish & Portuguese literature 880 Italian literature 890 Other literatures 900 History 900 History 910 Asia 920 Europe 930 Africa 940 North America 950 South America 960 Oceania and Polar regions 970 [Unassigned] 980 Geography 990 Biography === Expansion tables === Table 1. Standard subdivisions Table 2. Geographic Areas Table 3. Korean geographic areas Table 4. Korean historical period Table 5. Languages Table 6. Subdivisions of individual languages Table 7. Subdivisions of individual literatures Table 8. Subdivisions of individual religions == Usage == KDC is used by a wide range of libraries within Korea, including by the National Library of Korea and most school and public libraries in the country, along with some university libraries, such as the one at Keimyung University.

  Read more →
• 

  Interactive activation and competition networks

  Interactive activation and competition (IAC) networks are artificial neural networks used to model memory and intuitive generalizations. They are made up of nodes or artificial neurons which are arrayed and activated in ways that emulate the behaviors of human memory. The IAC model is used by the parallel distributed processing (PDP) Group and is associated with James L. McClelland and David E. Rumelhart; it is described in detail in their book Explorations in Parallel Distributed Processing: A Handbook of Models, Programs, and Exercises. This model does not contradict any currently known biological data or theories, and its performance is close enough to human performance as to warrant further investigation.

  Read more →
• 

  Rclone

  Rclone is an open source, multi threaded, command line computer program to manage or migrate content on cloud and other high latency storage. Its capabilities include sync, transfer, crypt, cache, union, compress and mount. The rclone website lists supported backends including S3 and Google Drive. Descriptions of rclone often carry the strapline "Rclone syncs your files to cloud storage". Those prior to 2020 include the alternative "Rsync for Cloud Storage". Rclone is well known for its rclone sync and rclone mount commands. It provides further management functions analogous to those ordinarily used for files on local disks, but which tolerate some intermittent and unreliable service. Rclone is commonly used with media servers such as Plex, Emby or Jellyfin to stream content direct from consumer file storage services. Official Ubuntu, Debian, Fedora, Gentoo, Arch, Brew, Chocolatey, and other package managers include rclone. == History == Nick Craig-Wood was inspired by rsync. Concerns about the noise and power costs arising from home computer servers prompted him to embrace cloud storage and he began developing rclone as open source software in 2012 under the name swiftsync. Rclone was promoted to stable version 1.00 in July 2014. In May 2017, Amazon Drive barred new users of rclone and other upload utilities, citing security concerns. Amazon Drive had been advertised as offering unlimited storage for £55 per year. Amazon's AWS S3 service continues to support new rclone users. The original rclone logo was updated in September 2018. In March 2020, Nick Craig-Wood resigned from Memset Ltd, a cloud hosting company he founded, to focus on open source software. Amazon's AWS April 2020 public sector blog explained how the Fred Hutch Cancer Research Center were using rclone in their Motuz tool to migrate very large biomedical research datasets in and out of AWS S3 object stores. In November 2020, rclone was updated to correct a weakness in the way it generated passwords. Passwords for encrypted remotes can be generated randomly by rclone or supplied by the user. In all versions of rclone from 1.49.0 to 1.53.2 the seed value for generated passwords was based on the number of seconds elapsed in the day, and therefore not truly random. CVE-2020-28924 recommended users upgrade to the latest version of rclone and check the passwords protecting their encrypted remotes. Release 1.55 of rclone in March 2021 included features sponsored by CERN and their CS3MESH4EOSC project. The work was EU funded to promote vendor-neutral application programming interfaces and protocols for synchronisation and sharing of academic data on cloud storage. == Backends and commands == Rclone supports the following services as backends. There are others, built on standard protocols such as WebDAV or S3, that work. WebDAV backends do not support rclone functionality dependent on server side checksum or modtime. Remotes are usually defined interactively from these backends, local disk, or memory (as S3), with rclone config. Rclone can further wrap those remotes with one or more of alias, chunk, compress, crypt or union, remotes. Once defined, the remotes are referenced by other rclone commands interchangeably with the local drive. Remote names are followed by a colon to distinguish them from local drives. For example, a remote example_remote containing a folder, or pseudofolder, myfolder is referred to within a command as a path example_remote:/myfolder. Rclone commands directly apply to remotes, or mount them for file access or streaming. With appropriate cache options the mount can be addressed as if a conventional, block level disk. Commands are provided to serve remotes over SFTP, HTTP, WebDAV, FTP and DLNA. Commands can have sub-commands and flags. Filters determine which files on a remote that rclone commands are applied to. rclone rc passes commands or new parameters to existing rclone sessions and has an experimental web browser interface. === Crypt remotes === Rclone's crypt implements encryption of files at rest in cloud storage. It layers an encrypted remote over a pre-existing, cloud or other remote. Crypt is commonly used to encrypt / decrypt media, for streaming, on consumer storage services such as Google Drive. Rclone's configuration file contains the crypt password. The password can be lightly obfuscated, or the whole rclone.conf file can be encrypted. Crypt can either encrypt file content and name, or additionally full paths. In the latter case there is a potential clash with encryption for cloud backends, such as Microsoft OneDrive, having limited path lengths. Crypt remotes do not encrypt object modification time or size. The encryption mechanism for content, name and path is available, for scrutiny, on the rclone website. Key derivation is with scrypt. === Example syntax (Linux) === These examples describe paths and file names but object keys behave similarly. To recursively copy files from directory remote_stuff, at the remote xmpl, to directory stuff in the home folder:- -v enables logging and -P, progress information. By default rclone checks the file integrity (hash) after copy; can retry each file up to three times if the operation is interrupted; uses up to four parallel transfer threads, and does not apply bandwidth throttling. Running the above command again copies any new or changed files at the remote to the local folder but, like default rsync behaviour, will not delete from the local directory, files which have been removed from the remote. To additionally delete files from the local folder which have been removed from the remote - more like the behaviour of rsync with a --delete flag:- And to delete files from the source after they have been transferred to the local directory - more like the behaviour of rsync with a --remove-source-file flag:- To mount the remote directory at a mountpoint in the pre-existing, empty stuff directory in the home directory (the ampersand at the end makes the mount command run as a background process):- Default rclone syntax can be modified. Alternative transfer, filter, conflict and backend specific flags are available. Performance choices include number of concurrent transfer threads; chunk size; bandwidth limit profiling, and cache aggression. == Academic evaluation == In 2018, University of Kentucky researchers published a conference paper comparing use of rclone and other command line, cloud data transfer agents for big data. The paper was published as a result of funding by the National Science Foundation. Later that year, University of Utah's Center for High Performance Computing examined the impact of rclone options on data transfer rates. == Rclone use at HPC research sites == Examples are University of Maryland, Iowa State University, Trinity College Dublin, NYU, BYU, Indiana University, CSC Finland, Utrecht University, University of Nebraska, University of Utah, North Carolina State University, Stony Brook, Tulane University, Washington State University, Georgia Tech, National Institutes of Health, Wharton, Yale, Harvard, Minnesota, Michigan State, Case Western Reserve University, University of South Dakota, Northern Arizona University, University of Pennsylvania, Stanford, University of Southern California, UC Santa Barbara, UC Irvine, UC Berkeley, and SURFnet. == Rclone and cybercrime == May 2020 reports stated rclone had been used by hackers to exploit Diebold Nixdorf ATMs with ProLock ransomware. The FBI issued a Flash Alert MI-000125-MW on May 4, 2020, in relation to the compromise. They issued a further, related alert 20200901–001 in September 2020. Attackers had exfiltrated / encrypted data from organisations involved in healthcare, construction, finance, and legal services. Multiple US government agencies, and industrial entities were affected. Researchers established the hackers spent about a month exploring the breached networks, using rclone to archive stolen data to cloud storage, before encrypting the target system. Reported targets included LaSalle County, and the city of Novi Sad. The FBI warned January 2021, in Private Industry Notification 20210106–001, of extortion activity using Egregor ransomware and rclone. Organisations worldwide had been threatened with public release of exfiltrated data. In some cases rclone had been disguised under the name svchost. Bookseller Barnes & Noble, US retailer Kmart, games developer Ubisoft and the Vancouver metro system have been reported as victims. An April 2021, cybersecurity investigation into SonicWall VPN zero-day vulnerability SNWLID-2021-0001 by FireEye's Mandiant team established attackers UNC2447 used rclone for reconnaissance and exfiltration of victims' files. Cybersecurity and Infrastructure Security Agency Analysis Report AR21-126A confirmed this use of rclone in FiveHands ransomware attacks. A June 2021, Microsoft Security Intelligence Twitter post identified use of rclone in BazaCall cyber attacks. The attackers sent emails e

  Read more →
• 

  TD-Gammon

  TD-Gammon is a computer backgammon program developed in the 1990s by Gerald Tesauro at IBM's Thomas J. Watson Research Center. Its name comes from the fact that it is an artificial neural net trained by a form of temporal-difference learning, specifically TD-Lambda. It explored strategies that humans had not pursued and led to advances in the theory of correct backgammon play. In 1993, TD-Gammon (version 2.1) was trained with 1.5 million games of self-play, and achieved a level of play just slightly below that of the top human backgammon players of the time. In 1998, during a 100-game series, it was defeated by the world champion by a mere margin of 8 points. Its unconventional assessment of some opening strategies had been accepted and adopted by expert players. TD-gammon is commonly cited as an early success of reinforcement learning and neural networks, and was cited in, for example, papers for deep Q-learning and AlphaGo. == Algorithm for play and learning == During play, TD-Gammon examines on each turn all possible legal moves and all their possible responses (lookahead search), feeds each resulting board position into its evaluation function, and chooses the move that leads to the board position that got the highest score. In this respect, TD-Gammon is no different than almost any other computer board-game program. TD-Gammon's innovation was in how it learned its evaluation function. TD-Gammon's learning algorithm consists of updating the weights in its neural net after each turn to reduce the difference between its evaluation of previous turns' board positions and its evaluation of the present turn's board position—hence "temporal-difference learning". The score of any board position is a set of four numbers reflecting the program's estimate of the likelihood of each possible game result: White wins normally, Black wins normally, White wins a gammon, Black wins a gammon. For the final board position of the game, the algorithm compares with the actual result of the game rather than its own evaluation of the board position. The core of TD-gammon is a neural network with 3 layers. The input layer has two types of neurons. One type codes for the board position. They are non-negative integers ranging from 0 to 15, indicating the number of White or Black checkers at each board location. There are 99 input neurons for each, totaling 198 neurons. Another type codes for hand-crafted features previously used in Neurogammon. These features encoded standard concepts used by human experts, such as "advanced anchor," "blockade strength," "home board strength" and the probability of a "blot" (single checker) being hit. The hidden layer contains hidden neurons. Later versions had more of these. The output layer contains 4 neurons, representing the network's estimate of the probability ("equity") that the current board would lead to. The 4 neurons code for: White normal win, White gammon win, Black normal win, Black gammon win. Backgammon win is so rare that Tesauro opted to not represent it. After each turn, the learning algorithm updates each weight in the neural net according to the following rule: w t + 1 − w t = α ( Y t + 1 − Y t ) ∑ k = 1 t λ t − k ∇ w Y k {\displaystyle w_{t+1}-w_{t}=\alpha (Y_{t+1}-Y_{t})\sum _{k=1}^{t}\lambda ^{t-k}\nabla _{w}Y_{k}} where: It was found that picking small λ {\displaystyle \lambda } offered performance roughly equally good, and large λ {\displaystyle \lambda } degraded performance. Because of this, after 1992, TD-Gammon was trained with λ = 0 {\displaystyle \lambda =0} , degenerating into standard TD-learning. This saved compute by a factor of 2. == Development history == Version 1.0 used simple 1-ply search: every next move is scored by the neural net, and the highest-scoring move is selected. Versions 2.0 and 2.1 used 2-ply search: Make a 1-ply analysis to remove unlikely moves ("forward pruning"). Make a 2-play minimax analysis for only the likely moves. Pick the best move, probability-weighted by each of the opponent's 21 possible dice rolls (weighting non-doubles twice as much as doubles). Versions 3.0 and 3.1 used 3-ply search, using 21 2 = 441 {\displaystyle 21^{2}=441} possible dice rolls instead of 21. The last version, 3.1, was trained specifically for an exhibition match against Malcolm Davis at the 1998 AAAI Hall of Champions. It lost at -8 points, mainly due to one blunder, where TD-Gammon opted to double and got gammoned at -32 points. == Experiments and stages of training == Unlike previous neural-net backgammon programs such as Neurogammon (also written by Tesauro), where an expert trained the program by supplying the "correct" evaluation of each position, TD-Gammon was at first programmed "knowledge-free". In early experimentation, using only a raw board encoding with no human-designed features, TD-Gammon reached a level of play comparable to Neurogammon: that of an intermediate-level human backgammon player. Even though TD-Gammon discovered insightful features on its own, Tesauro wondered if its play could be improved by using hand-designed features like Neurogammon's. Indeed, the self-training TD-Gammon with expert-designed features soon surpassed all previous computer backgammon programs. It stopped improving after about 1,500,000 games (self-play) using a three-layered neural network, with 198 input units encoding expert-designed features, 80 hidden units, and one output unit representing predicted probability of winning. == Advances in backgammon theory == TD-Gammon's exclusive training through self-play (rather than imitation learning) enabled it to explore strategies that humans previously had not considered or had ruled out erroneously. Its success with unorthodox strategies had a significant impact on the backgammon community. Late 1991, Bill Robertie, Paul Magriel, and Malcolm Davis, were invited to play against TD-Gammon (version 1.0). A total of 51 games were played, with TD-Gammon losing at -0.25 ppg. Robertie found TD-Gammon to be at the level of a competent advanced player, and better than any previous backgammon program. Robertie subsequently wrote about the use of TD-Gammon for backgammon study. For example, on the opening play, the conventional wisdom was that given a roll of 2-1, 4-1, or 5-1, White should move a single checker from point 6 to point 5. Known as "slotting", this technique trades the risk of a hit for the opportunity to develop an aggressive position. TD-Gammon found that the more conservative play of splitting 24-23 was superior. Tournament players began experimenting with TD-Gammon's move, and found success. Within a few years, slotting had disappeared from tournament play, replaced by splitting, though in 2006 it made a reappearance for 2-1. Backgammon expert Kit Woolsey found that TD-Gammon's positional judgement, especially its weighing of risk against safety, was superior to his own or any human's. TD-Gammon's excellent positional play was undercut by occasional poor endgame play. The endgame requires a more analytical approach, sometimes with extensive lookahead. TD-Gammon's limitation to two-ply lookahead put a ceiling on what it could achieve in this part of the game. TD-Gammon's strengths and weaknesses were the opposite of symbolic artificial intelligence programs and most computer software in general: it was good at matters that require an intuitive "feel" but bad at systematic analysis. It is also poor at doubling strategies. This is likely due to the fact that the neural network is trained without the doubling cube, with the doubling added by feeding the neural network's cubeless equity estimates into theoretically-based heuristic formulae. This was particularly the case in the 1998 exhibition match, where it played 100 games against Malcolm Davis. A single doubling blunder lost the match. TD-gammon was never commercialized or released to the public in some other form, but it inspired commercial backgammon programs based on neural networks, such as JellyFish (1994) and Snowie (1998).

  Read more →
• 

  STUDENT

  STUDENT is an early artificial intelligence program that solves algebra word problems. It is written in Lisp by Daniel G. Bobrow as his PhD thesis in 1964 (Bobrow 1964). It was designed to read and solve the kind of word problems found in high school algebra books. The program is often cited as an early accomplishment of AI in natural language processing. == Technical description == Within Project MAC at MIT, the STUDENT system was an early example of a question answering software, which uniquely involved natural language processing and symbolic programming. Other early attempts for solving algebra story problems were realized with 1960s hardware and software as well: for example, the Philips, Baseball and Synthex systems. STUDENT accepts an algebra story written in the English language as input, and generates a number as output. This is realized with a layered pipeline that consists of heuristics for pattern transformation. At first, sentences in English are converted into kernel sentences, which each contain a single piece of information. Next, the kernel sentences are converted into mathematical expressions. The knowledge base that supports the transformation contains 52 facts. STUDENT uses a rule-based system with logic inference. The rules are pre-programmed by the software developer and are able to parse natural language. More powerful techniques for natural language processing, such as machine learning, came into use later as hardware grew more capable, and gained popularity over simpler rule-based systems.

  Read more →