AI Assistant Grok

AI Assistant Grok — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

• 

  Computational humor

  Computational humor is a branch of computational linguistics and artificial intelligence which uses computers in humor research. It is a relatively new area, with the first dedicated conference organized in 1996. The first "computer model of a sense of humor" was suggested by Suslov as early as 1992. Investigation of the general scheme of the information processing show a possibility of a specific malfunction, conditioned by the necessity of a quick deletion from consciousness of a false version. This specific malfunction can be identified with a humorous effect on the psychological grounds; however, an essentially new ingredient, a role of timing, is added to a well known role of ambiguity. In biological systems, a sense of humour inevitably develops in the course of evolution, because its biological function consists in quickening the transmission of processed information into consciousness and in a more effective use of brain resources. A realization of this algorithm in neural networks explains naturally the mechanism of laughter: deletion of a false version corresponds to zeroing of some part of the neural network and excessive energy of neurons is thrown out to the motor cortex, arousing muscular contractions. Unfortunately, a practical realization of this algorithm needs extensive databases, whose creation in the automatic regime was suggested only recently . As a result, this magistral direction was not developed properly and subsequent investigations (see below) accepted somewhat specialized colouring. == Joke generators == === Pun generation === An approach to analysis of humor is classification of jokes. A further step is an attempt to generate jokes basing on the rules that underlie classification. Simple prototypes for computer pun generation were reported in the early 1990s, based on a natural language generator program, VINCI. Graeme Ritchie and Kim Binsted in their 1994 research paper described a computer program, JAPE, designed to generate question-answer-type puns from a general, i.e., non-humorous, lexicon. (The program name is an acronym for "Joke Analysis and Production Engine".) Some examples produced by JAPE are: Q: What is the difference between leaves and a car? A: One you brush and rake, the other you rush and brake. Q: What do you call a strange market? A: A bizarre bazaar. Since then the approach has been improved, and the latest report, dated 2007, describes the STANDUP joke generator, implemented in the Java programming language. The STANDUP generator was tested on children within the framework of analyzing its usability for language skills development for children with communication disabilities, e.g., because of cerebral palsy. (The project name is an acronym for "System To Augment Non-speakers' Dialog Using Puns" and an allusion to standup comedy.) Children responded to this "language playground" with enthusiasm, and showed marked improvement on certain types of language tests. The two young people, who used the system over a ten-week period, regaled their peers, staff, family and neighbors with jokes such as: "What do you call a spicy missile? A hot shot!" Their joy and enthusiasm at entertaining others was inspirational. === Other === Stock and Strapparava described a program to generate funny acronyms. == Joke recognition == A statistical machine learning algorithm to detect whether a sentence contained a "That's what she said" double entendre was developed by Kiddon and Brun (2011). There is an open-source Python implementation of Kiddon & Brun's TWSS system. A program to recognize knock-knock jokes was reported by Taylor and Mazlack. This kind of research is important in analysis of human–computer interaction. An application of machine learning techniques for the distinguishing of joke texts from non-jokes was described by Mihalcea and Strapparava (2006). Takizawa et al. (1996) reported on a heuristic program for detecting puns in the Japanese language. == Applications == A possible application for assistance in language acquisition is described in the section "Pun generation". Another envisioned use of joke generators is in cases of a steady supply of jokes where quantity is more important than quality. Another obvious, yet remote, direction is automated joke appreciation. It is known that humans interact with computers in ways similar to interacting with other humans that may be described in terms of personality, politeness, flattery, and in-group favoritism. Therefore, the role of humor in human–computer interaction is being investigated. In particular, humor generation in user interface to ease communications with computers was suggested. Craig McDonough implemented the Mnemonic Sentence Generator, which converts passwords into humorous sentences. Based on the incongruity theory of humor, it is suggested that the resulting meaningless but funny sentences are easier to remember. For example, the password AjQA3Jtv is converted into "Arafat joined Quayle's Ant, while TARAR Jeopardized thurmond's vase," an example chosen by combining politicians names with verbs and common nouns. == Related research == John Allen Paulos is known for his interest in mathematical foundations of humor. His book Mathematics and Humor: A Study of the Logic of Humor demonstrates structures common to humor and formal sciences (mathematics, linguistics) and develops a mathematical model of jokes based on catastrophe theory. Conversational systems which have been designed to take part in Turing test competitions generally have the ability to learn humorous anecdotes and jokes. Because many people regard humor as something particular to humans, its appearance in conversation can be quite useful in convincing a human interrogator that a hidden entity, which could be a machine or a human, is in fact a human.

  Read more →
• 

  Alexey Chervonenkis

  Alexey Yakovlevich Chervonenkis (Russian: Алексей Яковлевич Червоненкис; 7 September 1938 – 22 September 2014) was a Soviet and Russian mathematician. Along with Vladimir Vapnik, he was one of the main developers of the Vapnik–Chervonenkis theory, also known as the "fundamental theory of learning", an important part of computational learning theory. Chervonenkis held joint appointments with the Russian Academy of Sciences and Royal Holloway, University of London. Alexey Chervonenkis got lost in Losiny Ostrov National Park on 22 September 2014, and later during a search operation was found dead near Mytishchi, a suburb of Moscow. He had died of hypothermia.

  Read more →
• 

  AlphaGo Zero

  AlphaGo Zero is a version of DeepMind's Go software AlphaGo. AlphaGo's team published an article in Nature in October 2017 introducing AlphaGo Zero, a version created without using data from human games, and stronger than any previous version. By playing games against itself, AlphaGo Zero: surpassed the strength of AlphaGo Lee in three days by winning 100 games to 0; reached the level of AlphaGo Master in 21 days; and exceeded all previous versions in 40 days. Training artificial intelligence (AI) without datasets derived from human experts has significant implications for the development of AI with superhuman skills, as expert data is "often expensive, unreliable, or simply unavailable." Demis Hassabis, the co-founder and CEO of DeepMind, said that AlphaGo Zero was so powerful because it was "no longer constrained by the limits of human knowledge". Furthermore, AlphaGo Zero performed better than standard deep reinforcement learning models (such as Deep Q-Network implementations) due to its integration of Monte Carlo tree search. David Silver, one of the first authors of DeepMind's papers published in Nature on AlphaGo, said that it is possible to have generalized AI algorithms by removing the need to learn from humans. Google later developed AlphaZero, a generalized version of AlphaGo Zero that could play chess and shōgi in addition to Go. In December 2017, AlphaZero beat the 3-day version of AlphaGo Zero by winning 60 games to 40, and with 8 hours of training it outperformed AlphaGo Lee on an Elo scale. AlphaZero also defeated a top chess program (Stockfish) and a top Shōgi program (Elmo). == Architecture == The network in AlphaGo Zero is a ResNet with two heads. The stem of the network takes as input a 17x19x19 tensor representation of the Go board. 8 channels are the positions of the current player's stones from the last eight time steps. (1 if there is a stone, 0 otherwise. If the time step go before the beginning of the game, then 0 in all positions.) 8 channels are the positions of the other player's stones from the last eight time steps. 1 channel is all 1 if black is to move, and 0 otherwise. The body is a ResNet with either 20 or 40 residual blocks and 256 channels. There are two heads, a policy head and a value head. Policy head outputs a logit array of size 19 × 19 + 1 {\displaystyle 19\times 19+1} , representing the logit of making a move in one of the points, plus the logit of passing. Value head outputs a number in the range ( − 1 , + 1 ) {\displaystyle (-1,+1)} , representing the expected score for the current player. -1 represents current player losing, and +1 winning. == Training == AlphaGo Zero's neural network was trained using TensorFlow, with 64 GPU workers and 19 CPU parameter servers. Only four TPUs were used for inference. The neural network initially knew nothing about Go beyond the rules. Unlike earlier versions of AlphaGo, Zero only perceived the board's stones, rather than having some rare human-programmed edge cases to help recognize unusual Go board positions. The AI engaged in reinforcement learning, playing against itself until it could anticipate its own moves and how those moves would affect the game's outcome. In the first three days AlphaGo Zero played 4.9 million games against itself in quick succession. It appeared to develop the skills required to beat top humans within just a few days, whereas the earlier AlphaGo took months of training to achieve the same level. According to Epoch.ai, training cost 3e23 FLOPs. For comparison, the researchers also trained a version of AlphaGo Zero using human games, AlphaGo Master, and found that it learned more quickly, but actually performed more poorly in the long run. DeepMind submitted its initial findings in a paper to Nature in April 2017, which was then published in October 2017. == Hardware cost == The hardware cost for a single AlphaGo Zero system in 2017, including the four TPUs, has been quoted as around $25 million. == Applications == According to Hassabis, AlphaGo's algorithms are likely to be of the most benefit to domains that require an intelligent search through an enormous space of possibilities, such as protein folding (see AlphaFold) or accurately simulating chemical reactions. AlphaGo's techniques are probably less useful in domains that are difficult to simulate, such as learning how to drive a car. DeepMind stated in October 2017 that it had already started active work on attempting to use AlphaGo Zero technology for protein folding, and stated it would soon publish new findings. == Reception == AlphaGo Zero was widely regarded as a significant advance, even when compared with its groundbreaking predecessor, AlphaGo. Oren Etzioni of the Allen Institute for Artificial Intelligence called AlphaGo Zero "a very impressive technical result" in "both their ability to do it—and their ability to train the system in 40 days, on four TPUs". The Guardian called it a "major breakthrough for artificial intelligence", citing Eleni Vasilaki of Sheffield University and Tom Mitchell of Carnegie Mellon University, who called it an impressive feat and an “outstanding engineering accomplishment" respectively. Mark Pesce of the University of Sydney called AlphaGo Zero "a big technological advance" taking us into "undiscovered territory". Gary Marcus, a psychologist at New York University, has cautioned that for all we know, AlphaGo may contain "implicit knowledge that the programmers have about how to construct machines to play problems like Go" and will need to be tested in other domains before being sure that its base architecture is effective at much more than playing Go. In contrast, DeepMind is "confident that this approach is generalisable to a large number of domains". In response to the reports, South Korean Go professional Lee Sedol said, "The previous version of AlphaGo wasn’t perfect, and I believe that’s why AlphaGo Zero was made." On the potential for AlphaGo's development, Lee said he will have to wait and see but also said it will affect young Go players. Mok Jin-seok, who directs the South Korean national Go team, said the Go world has already been imitating the playing styles of previous versions of AlphaGo and creating new ideas from them, and he is hopeful that new ideas will come out from AlphaGo Zero. Mok also added that general trends in the Go world are now being influenced by AlphaGo's playing style. "At first, it was hard to understand and I almost felt like I was playing against an alien. However, having had a great amount of experience, I’ve become used to it," Mok said. "We are now past the point where we debate the gap between the capability of AlphaGo and humans. It’s now between computers." Mok has reportedly already begun analyzing the playing style of AlphaGo Zero along with players from the national team. "Though having watched only a few matches, we received the impression that AlphaGo Zero plays more like a human than its predecessors," Mok said. Chinese Go professional Ke Jie commented on the remarkable accomplishments of the new program: "A pure self-learning AlphaGo is the strongest. Humans seem redundant in front of its self-improvement." == Comparison with predecessors == == AlphaZero == On 5 December 2017, DeepMind team released a preprint on arXiv, introducing AlphaZero, a program using generalized AlphaGo Zero's approach, which achieved within 24 hours a superhuman level of play in chess, shogi, and Go, defeating world-champion programs, Stockfish, Elmo, and 3-day version of AlphaGo Zero in each case. AlphaZero (AZ) is a more generalized variant of the AlphaGo Zero (AGZ) algorithm, and is able to play shogi and chess as well as Go. Differences between AZ and AGZ include: AZ has hard-coded rules for setting search hyperparameters. The neural network is now updated continually. Chess (unlike Go) can end in a tie; therefore AZ can take into account the possibility of a tie game. An open source program, Leela Zero, based on the ideas from the AlphaGo papers is available. It uses a GPU instead of the TPUs recent versions of AlphaGo rely on.

  Read more →
• 

  WYSIWYM (interaction technique)

  What you see is what you meant (WYSIWYM) is a text editing interaction technique that emerged from two projects at University of Brighton. It allows users to create abstract knowledge representations such as those required by the Semantic Web using a natural language interface. Natural language understanding (NLU) technology is not employed. Instead, natural language generation (NLG) is used in a highly interactive manner. The text editor accepts repeated refinement of a selected span of text as it becomes progressively less vacuous of authored semantics. Using a mouse, a text property held in the evolving text can be further refined by a set of options derived by NLG from a built-in ontology. An invisible representation of the semantic knowledge is created which can be used for multilingual document generation, formal knowledge formation, or any other task that requires formally specified information. The two projects at Brighton worked in the field of Conceptual Authoring to lay a foundation for further research and development of a Semantic Web Authoring Tool (SWAT). This tool has been further explored as a means for developing a knowledge base by those without prior experience with Controlled Natural Language tools.

  Read more →
• 

  Point distribution model

  The point distribution model is a model for representing the mean geometry of a shape and some statistical modes of geometric variation inferred from a training set of shapes. == Background == The point distribution model concept has been developed by Cootes, Taylor et al. and became a standard in computer vision for the statistical study of shape and for segmentation of medical images where shape priors really help interpretation of noisy and low-contrasted pixels/voxels. The latter point leads to active shape models (ASM) and active appearance models (AAM). Point distribution models rely on landmark points. A landmark is an annotating point posed by an anatomist onto a given locus for every shape instance across the training set population. For instance, the same landmark will designate the tip of the index finger in a training set of 2D hands outlines. Principal component analysis (PCA), for instance, is a relevant tool for studying correlations of movement between groups of landmarks among the training set population. Typically, it might detect that all the landmarks located along the same finger move exactly together across the training set examples showing different finger spacing for a flat-posed hands collection. == Details == First, a set of training images are manually landmarked with enough corresponding landmarks to sufficiently approximate the geometry of the original shapes. These landmarks are aligned using the generalized procrustes analysis, which minimizes the least squared error between the points. k {\displaystyle k} aligned landmarks in two dimensions are given as X = ( x 1 , y 1 , … , x k , y k ) {\displaystyle \mathbf {X} =(x_{1},y_{1},\ldots ,x_{k},y_{k})} . It's important to note that each landmark i ∈ { 1 , … k } {\displaystyle i\in \lbrace 1,\ldots k\rbrace } should represent the same anatomical location. For example, landmark #3, ( x 3 , y 3 ) {\displaystyle (x_{3},y_{3})} might represent the tip of the ring finger across all training images. Now the shape outlines are reduced to sequences of k {\displaystyle k} landmarks, so that a given training shape is defined as the vector X ∈ R 2 k {\displaystyle \mathbf {X} \in \mathbb {R} ^{2k}} . Assuming the scattering is gaussian in this space, PCA is used to compute normalized eigenvectors and eigenvalues of the covariance matrix across all training shapes. The matrix of the top d {\displaystyle d} eigenvectors is given as P ∈ R 2 k × d {\displaystyle \mathbf {P} \in \mathbb {R} ^{2k\times d}} , and each eigenvector describes a principal mode of variation along the set. Finally, a linear combination of the eigenvectors is used to define a new shape X ′ {\displaystyle \mathbf {X} '} , mathematically defined as: X ′ = X ¯ + P b {\displaystyle \mathbf {X} '={\overline {\mathbf {X} }}+\mathbf {P} \mathbf {b} } where X ¯ {\displaystyle {\overline {\mathbf {X} }}} is defined as the mean shape across all training images, and b {\displaystyle \mathbf {b} } is a vector of scaling values for each principal component. Therefore, by modifying the variable b {\displaystyle \mathbf {b} } an infinite number of shapes can be defined. To ensure that the new shapes are all within the variation seen in the training set, it is common to only allow each element of b {\displaystyle \mathbf {b} } to be within ± {\displaystyle \pm } 3 standard deviations, where the standard deviation of a given principal component is defined as the square root of its corresponding eigenvalue. PDM's can be extended to any arbitrary number of dimensions, but are typically used in 2D image and 3D volume applications (where each landmark point is R 2 {\displaystyle \mathbb {R} ^{2}} or R 3 {\displaystyle \mathbb {R} ^{3}} ). == Discussion == An eigenvector, interpreted in euclidean space, can be seen as a sequence of k {\displaystyle k} euclidean vectors associated to corresponding landmark and designating a compound move for the whole shape. Global nonlinear variation is usually well handled provided nonlinear variation is kept to a reasonable level. Typically, a twisting nematode worm is used as an example in the teaching of kernel PCA-based methods. Due to the PCA properties: eigenvectors are mutually orthogonal, form a basis of the training set cloud in the shape space, and cross at the 0 in this space, which represents the mean shape. Also, PCA is a traditional way of fitting a closed ellipsoid to a Gaussian cloud of points (whatever their dimension): this suggests the concept of bounded variation. The idea behind PDMs is that eigenvectors can be linearly combined to create an infinity of new shape instances that will 'look like' the one in the training set. The coefficients are bounded alike the values of the corresponding eigenvalues, so as to ensure the generated 2n/3n-dimensional dot will remain into the hyper-ellipsoidal allowed domain—allowable shape domain (ASD).

  Read more →
• 

  Tree (abstract data type)

  In computer science, a tree is a widely used abstract data type that represents a hierarchical tree structure with a set of connected nodes. Each node in the tree can be connected to many children (depending on the type of tree), but must be connected to exactly one parent, except for the root node, which has no parent (i.e., the root node as the top-most node in the tree hierarchy). These constraints mean there are no cycles or "loops" (no node can be its own ancestor), and also that each child can be treated like the root node of its own subtree, making recursion a useful technique for tree traversal. In contrast to linear data structures, many trees cannot be represented by relationships between neighboring nodes (parent and children nodes of a node under consideration, if they exist) in a single straight line (called edge or link between two adjacent nodes). Binary trees are a commonly used type, which constrain the number of children for each parent to at most two. When the order of the children is specified, this data structure corresponds to an ordered tree in graph theory. A value or pointer to other data may be associated with every node in the tree, or sometimes only with the leaf nodes, which have no children nodes. The abstract data type (ADT) can be represented in a number of ways, including a list of parents with pointers to children, a list of children with pointers to parents, or a list of nodes and a separate list of parent-child relations (a specific type of adjacency list). Representations might also be more complicated, for example using indexes or ancestor lists for performance. Trees as used in computing are similar to but can be different from mathematical constructs of trees in graph theory, trees in set theory, and trees in descriptive set theory. == Terminology == A node is a structure which may contain data and connections to other nodes, sometimes called edges or links. Each node in a tree has zero or more child nodes, which are below it in the tree (by convention, trees are drawn with descendants going downwards). A node that has a child is called the child's parent node (or superior). All nodes have exactly one parent, except the topmost root node, which has none. A node might have many ancestor nodes, such as the parent's parent. Child nodes with the same parent are sibling nodes. Typically siblings have an order, with the first one conventionally drawn on the left. Some definitions allow a tree to have no nodes at all, in which case it is called empty. An internal node (also known as an inner node, inode for short, or branch node) is any node of a tree that has child nodes. Similarly, an external node (also known as an outer node, leaf node, or terminal node) is any node that does not have child nodes. The height of a node is the length of the longest downward path to a leaf from that node. The height of the root is the height of the tree. The depth of a node is the length of the path to its root (i.e., its root path). Thus the root node has depth zero, leaf nodes have height zero, and a tree with only a single node (hence both a root and leaf) has depth and height zero. Conventionally, an empty tree (tree with no nodes, if such are allowed) has height −1. Each non-root node can be treated as the root node of its own subtree, which includes that node and all its descendants. Other terms used with trees: Neighbor Parent or child. Ancestor A node reachable by repeated proceeding from child to parent. Descendant A node reachable by repeated proceeding from parent to child. Also known as subchild. Degree For a given node, its number of children. A leaf, by definition, has degree zero. Degree of tree The degree of a tree is the maximum degree of a node in the tree. Distance The number of edges along the shortest path between two nodes. Level The level of a node is the number of edges along the unique path between it and the root node. This is the same as depth. Width The number of nodes in a level. Breadth The number of leaves. Complete tree A tree with every level filled, except the last. Forest A set of one or more disjoint trees. Ordered tree A rooted tree in which an ordering is specified for the children of each vertex. Size of a tree Number of nodes in the tree. == Common operations == Enumerating all the items Enumerating a section of a tree Searching for an item Adding a new item at a certain position on the tree Deleting an item Pruning: Removing a whole section of a tree Grafting: Adding a whole section to a tree Finding the root for any node Finding the lowest common ancestor of two nodes === Traversal and search methods === Stepping through the items of a tree, by means of the connections between parents and children, is called walking the tree, and the action is a walk of the tree. Often, an operation might be performed when a pointer arrives at a particular node. A walk in which each parent node is traversed before its children is called a pre-order walk; a walk in which the children are traversed before their respective parents are traversed is called a post-order walk; a walk in which a node's left subtree, then the node itself, and finally its right subtree are traversed is called an in-order traversal. (This last scenario, referring to exactly two subtrees, a left subtree and a right subtree, assumes specifically a binary tree.) A level-order walk effectively performs a breadth-first search over the entirety of a tree; nodes are traversed level by level, where the root node is visited first, followed by its direct child nodes and their siblings, followed by its grandchild nodes and their siblings, etc., until all nodes in the tree have been traversed. == Representations == There are many different ways to represent trees. In working memory, nodes are typically dynamically allocated records with pointers to their children, their parents, or both, as well as any associated data. If of a fixed size, the nodes might be stored in a list. Nodes and relationships between nodes might be stored in a separate special type of adjacency list. In relational databases, nodes are typically represented as table rows, with indexed row IDs facilitating pointers between parents and children. Nodes can also be stored as items in an array, with relationships between them determined by their positions in the array (as in a binary heap). A binary tree can be implemented as a list of lists: the head of a list (the value of the first term) is the left child (subtree), while the tail (the list of second and subsequent terms) is the right child (subtree). This can be modified to allow values as well, as in Lisp S-expressions, where the head (value of first term) is the value of the node, the head of the tail (value of second term) is the left child, and the tail of the tail (list of third and subsequent terms) is the right child. Ordered trees can be naturally encoded by finite sequences, for example with natural numbers. == Examples of trees and non-trees == == Type theory == As an abstract data type, the abstract tree type T with values of some type E is defined, using the abstract forest type F (list of trees), by the functions: value: T → E children: T → F nil: () → F node: E × F → T with the axioms: value(node(e, f)) = e children(node(e, f)) = f In terms of type theory, a tree is an inductive type defined by the constructors nil (empty forest) and node (tree with root node with given value and children). == Mathematical terminology == Viewed as a whole, a tree data structure is an ordered tree, generally with values attached to each node. Concretely, it is (if required to be non-empty): A rooted tree with the "away from root" direction (a more narrow term is an "arborescence"), meaning: A directed graph, whose underlying undirected graph is a tree (any two vertices are connected by exactly one simple path), with a distinguished root (one vertex is designated as the root), which determines the direction on the edges (arrows point away from the root; given an edge, the node that the edge points from is called the parent and the node that the edge points to is called the child), together with: an ordering on the child nodes of a given node, and a value (of some data type) at each node. Often trees have a fixed (more properly, bounded) branching factor (outdegree), particularly always having two child nodes (possibly empty, hence at most two non-empty child nodes), hence a "binary tree". Allowing empty trees makes some definitions simpler, some more complicated: a rooted tree must be non-empty, hence if empty trees are allowed the above definition instead becomes "an empty tree or a rooted tree such that ...". On the other hand, empty trees simplify defining fixed branching factor: with empty trees allowed, a binary tree is a tree such that every node has exactly two children, each of which is a tree (possibly empty). == Applications == Trees are commonly used to represent or manipulate hierarchical data in ap

  Read more →
• 

  Ishikawa diagram

  Ishikawa diagrams (also called fishbone diagrams, herringbone diagrams, cause-and-effect diagrams) are causal diagrams created by Kaoru Ishikawa that show the potential causes of a specific event. Common uses of the Ishikawa diagram are product design and quality defect prevention to identify potential factors causing an overall effect. Each cause or reason for imperfection is a source of variation. Causes are usually grouped into major categories to identify and classify these sources of variation. == Overview == The defect, or the problem to be solved, is shown as the fish's head, facing to the right, with the causes extending to the left as fishbones; the ribs branch off the backbone for major causes, with sub-branches for root-causes, to as many levels as required. Ishikawa diagrams were popularized in the 1960s by Kaoru Ishikawa, who pioneered quality management processes in the Kawasaki shipyards, and in the process became one of the founding fathers of modern management. The basic concept was first used in the 1920s, and is considered one of the seven basic tools of quality control. It is known as a fishbone diagram because of its shape, similar to the side view of a fish skeleton. Mazda Motors famously used an Ishikawa diagram in the development of the Miata (MX5) sports car. == Root causes == Root-cause analysis is intended to reveal key relationships among various variables, and the possible causes provide additional insight into process behavior. It shows high-level causes that lead to the problem encountered by providing a snapshot of the current situation. There can be confusion about the relationships between problems, causes, symptoms and effects. Smith highlights this and the common question “Is that a problem or a symptom?” which mistakenly presumes that problems and symptoms are mutually exclusive categories. A problem is a situation that bears improvement; a symptom is the effect of a cause: a situation can be both a problem and a symptom. At a practical level, a cause is whatever is responsible for, or explains, an effect - a factor "whose presence makes a critical difference to the occurrence of an outcome". The causes emerge by analysis, often through brainstorming sessions, and are grouped into categories on the main branches off the fishbone. To help structure the approach, the categories are often selected from one of the common models shown below, but may emerge as something unique to the application in a specific case. Each potential cause is traced back to find the root cause, often using the 5 Whys technique. Typical categories include: === The 5 Ms (used in manufacturing) === Originating with lean manufacturing and the Toyota Production System, the 5 Ms is one of the most common frameworks for root-cause analysis: Manpower / Mindpower (physical or knowledge work, includes: kaizens, suggestions) Machine (equipment, technology) Material (includes raw material, consumables, and information) Method (process) Measurement / medium (inspection, environment) These have been expanded by some to include an additional three, and are referred to as the 8 Ms: Mission / mother nature (purpose, environment) Management / money power (leadership) Maintenance === The 8 Ps (used in product marketing) === This common model for identifying crucial attributes for planning in product marketing is often also used in root-cause analysis as categories for the Ishikawa diagram: Product (or service) Price Place Promotion People (personnel) Process Physical evidence (proof) Performance === The 4 or 5 Ss (used in service industries) === An alternative used for service industries, uses four categories of possible cause: Surroundings: Refers to the environment in which the process occurs. Suppliers: Refers to external parties that provide inputs—raw materials, components, or services. Systems: Refers to the procedures, processes, and technologies used to perform the work. Skill: Refers to the human factor, particularly the knowledge and abilities of employees. Safety: Refers to physical and psychological well-being in the workplace. == Use in specific industries == The Ishikawa diagram has been widely adopted across various industries as an effective tool for root cause analysis in quality, efficiency, and safety-related issues. Its versatility allows it to be applied in both manufacturing and service contexts. In the manufacturing industry, particularly in the automotive and electronics sectors, the diagram is frequently used in continuous improvement initiatives such as Six Sigma and Lean Manufacturing. Quality teams use it to identify causes related to materials, methods, machinery, manpower, environment, and measurement, facilitating informed decision-making to reduce defects and optimize processes. In the food industry, the Ishikawa diagram is applied to analyze issues related to food safety, temperature control, cross-contamination, and regulatory compliance. Its use enables companies to identify improvement opportunities in production, packaging, and distribution stages. In the pharmaceutical sector, it is a key tool in process validation, quality control, and compliance with Good Manufacturing Practices (GMP). It helps visualize factors affecting product quality from formulation to storage. It has also been successfully implemented in sectors such as aerospace, pulp and paper, construction, education, and healthcare, where it supports structured problem-solving and promotes continuous improvement and a culture of quality.

  Read more →
• 

  Transparency in Frontier Artificial Intelligence Act

  The Transparency in Frontier Artificial Intelligence Act, also referred to as SB-53, is a 2025 California law which mandates increased transparency for companies building artificial intelligence. SB-53 is primarily focused on assessing and reducing potential catastrophic risks from AI, and is the first bill addressing such risks to be passed into law in America. The bill requires companies to create publicly accessible documents assessing potential "catastrophic risk[s]" from their AI models, as well as publishing documentation on how the model incorporates national and international safety standards. SB-53 also sets up whistleblower protections and procedures for alerting the government to a "critical safety incident". == History == SB-53 was preceded in 2024 by the unsuccessful Safe and Secure Innovation for Frontier Artificial Intelligence Models Act ("SB-1047"), a proposed bill authored by Senator Scott Wiener which was vetoed by Governor Gavin Newsom. Afterwords, Newsom created a "Joint California AI Policy Working Group" to provide recommendations for AI regulation, which guided the drafting of SB-53. Senator Scott Wiener introduced the bill on January 7, 2025, and after a series of amendments, SB-53 passed the Senate 29-8 on September 13. Governor Gavin Newsom approved the bill on September 25, passing it into law. == Provisions == SB-53 applies primarily to companies making at least $500 million in yearly gross revenue. It defines a “frontier model” as any AI trained with over 1026 FLOPS (including fine-tuning), including unreleased internal models. Both the financial and computational thresholds must be met before most of the law is applied, although the threshold can be lowered or otherwise updated by the California Department of Technology in an annual review starting in 2027. Most of the bill's provisions are focused on "catastrophic risks" from AI, which are defined as incidents in which a model contributes to more than 50 deaths or serious injuries, or causes more than one billion dollars ($1,000,000,000) in economic damage from AI-assisted acts (such as cyberattacks or the creation of biological weapons). The bill requires companies to provide publicly accessible safety frameworks for frontier AI models, describing how the company tests for catastrophic risk from its AI, and how it implements protections against such risks. This includes addressing the possibility that the AI may attempt to circumvent internal guardrails or oversight mechanisms. (Certain safety incidents, such as dangerously deceptive model behavior, physical injury, or death, must be reported to California Office of Emergency Services (OES) within 15 days, unless the incident poses imminent physical risk, in which case it must be reported immediately.) The company must follow its published framework, and if any changes are made, the framework should be updated within 30 days, and justification for said changes must also be made public. Additionally, all frontier companies are required to publish basic information about newly released frontier models (such as terms of service, supported languages, and intended use), although only large companies (making over $500 million annually) need to publish full safety frameworks. SB-53 also establishes various whistleblower protections for covered employees. Large companies must have anonymous whistleblowing channels in place which protect employees from retaliation from reporting risks to state or federal authorities if they have reasonable cause to believe that their employer is substantially risking public health and safety.

  Read more →
• 

  NetMiner

  NetMiner is an all-in-one software platform for analyzing and visualizing complex network data, based on Social Network Analysis (SNA). Originally released in 2001, it supports research and education in a wide range of domains through interactive and visual data exploration. This tool allows researchers to explore their network data visually and interactively, and helps them to detect underlying patterns and structures of the network. It has also been recognized for its comprehensive features and user-friendly interface in comparative reviews of SNA software packages. == Features == === Integrated Data Environment === NetMiner supports unified management of diverse data types—including network (nodes and links), tabular, and unstructured text data—within a single platform. This enables users to perform the entire analysis workflow seamlessly without switching between tools. NetMiner also supports a wide range of analytical methods, allowing users to derive new insights by combining multiple approaches. Analytical results can be saved and reused across workflows(Add to Dataset) Graph and Network Analysis: Includes Centrality, Community Detection, Blockmodeling, and Similarity Measures. Machine learning: Provides algorithms for regression, classification, clustering, ensemble modeling and XAI(Explainable AI) Graph Neural Networks (GNNs): Supports models such as GraphSAGE, GCN, and GAT to learn from both node attributes and graph structure. Natural language processing (NLP): Uses pretrained deep learning models to analyze unstructured text, including named entity recognition and keyword extraction. Text mining and Text network analysis: Supports construction of word co-occurrence networks and topic modeling using LDA, BERTopic, enabling identification of thematic patterns and semantic structures in text data. Data Visualization: Offers advanced network visualization features, supporting multiple layout algorithms. Analytical outcomes such as centrality or community detection can be directly reflected in the network map via node size, color, and position, enhancing intuitive understanding. === AI Assistant === NetMiner integrates with external large language models such as OpenAI GPT and Google Gemini to interpret complex analysis results in natural language, summarize key findings, and suggest next steps for exploration. === Workflow and Usability === Designed to follow the structure of real-world data analysis workflows, NetMiner adopts a hierarchical data organization (Project → Workspace → Dataset → Data Item). Its web-based user interface improves clarity and reduces complexity. NetMiner 5 supports Windows 10 or higher and macOS 11 or later with M1 chip. Both academic and commercial licenses are available. == Extension == NetMiner Extension is small program to extend the functionality of NetMiner. In other words, it enables you to customize NetMiner according to your needs. By adding ‘NetMiner Extension’, you can expand your research. === Web Data Collection === NetMiner allows users to collect data from services such as YouTube, OpenAlex, Springer, and KCI via Open APIs. Collected data is automatically preprocessed and transformed to fit NetMiner’s internal structure, requiring no additional coding or external tools. SNS Data Collector: It collects social media data from YouTube, which has a large number of social media users worldwide. Biblio Data Collector: It collects the bibliographic data from Springer, OpenAlex, and KCI essential for research trend analysis. == File formats == === NetMiner data file format === .NMF === Importable/exportable formats === Plain text data: .TXT, .CSV Microsoft Excel data: .XLS, .XLSX Unstructured text data: .TXT, .CSV, .XLS(X) ※ NetMiner 4 only NetMiner 2 data: .NTF UCINet data: .DL, .DAT Pajek data: .NET, .VEC, .CLU, .PER StOCNET data file: .DAT Graph Modelling Language data: .GML(importing only) Related software UCINET Pajek Gephi StoCNET == Data structure == === Hierarchy of NetMiner data structure === NetMiner 5 supports not only graph data composed of nodes and links, but also tabular and unstructured data without fixed schema or identifiers. This enables users to easily import a wide variety of raw and unstructured data suitable for machine learning applications. Within a single workspace, users can manage node sets, link sets, and structured/unstructured data simultaneously. Multiple graph layers under a node set can be organized in a tree structure, allowing for intuitive understanding of the data currently being analyzed. == Release history == The first version of NetMiner was released on Dec 21, 2001. There have been five major updates from 2001. === NetMiner 5 === Released on June 9, 2025. NetMiner 5 retains the core features and no-code concept of NetMiner 4, but has evolved by integrating cutting-edge AI technologies. AI Assistant, Personal Analytics Tutor Support for Graph, Structured, and Unstructured Data Graph Analytics / Social Network Analysis Machine Learning(M/L) & XAI Graph Machine Learning(GML): Graph Neural Network Text Mining: Natural Language Processing(NLP), Text Network, Topic Modeling Data Visualization === NetMiner 4 (2011) === Latest version is 4.5.1. Introduced Python scripting, encrypted NMF format, semantic analysis tools (word cloud, topic modeling), and Extension - Data Collector. === NetMiner 3 (2007) === Enhanced scalability, integrated analysis-visualization modules, and DB import from Oracle, MS SQL. === NetMiner 2 (2003) === Improved statistical and network measures, visualization algorithms, and external data import modules.

  Read more →
• 

  National Library of Medicine classification

  The National Library of Medicine (NLM) classification system is a library indexing system covering the fields of medicine and preclinical basic sciences. Operated and maintained by the U.S. National Library of Medicine, the NLM classification is patterned after the Library of Congress (LC) Classification system: alphabetical letters denote broad subject categories which are subdivided by numbers. For example, QW 279 would indicate a book on an aspect of microbiology or immunology. The one- or two-letter alphabetical codes in the NLM classification use a limited range of letters: only QS–QZ and W–WZ. This allows the NLM system to co-exist with the larger LC coding scheme as neither of these ranges are used in the LC system. There are, however, three pre-existing codes in the LC system which overlap with the NLM: Human Anatomy (QM), Microbiology (QR), and Medicine (R). To avoid further confusion, these three codes are not used in the NLM. The headings for the individual schedules (letters or letter pairs) are given in brief form (e.g., QW - Microbiology and Immunology; WG - Cardiovascular System) and together they provide an outline of the subjects covered by the NLM classification. Headings are interpreted broadly and include the physiological system, the specialties connected with them, the regions of the body chiefly concerned and subordinate related fields. The NLM system is hierarchical, and within each schedule, division by organ usually has priority. Each main schedule, as well as some sub-sections, begins with a group of form numbers ranging generally from 1–49 which classify materials by publication type, e.g., dictionaries, atlases, laboratory manuals, etc. The main schedules QS-QZ, W-WY, and WZ (excluding the range WZ 220–270) classify works published after 1913; the 19th century schedule is used for works published 1801–1913; and WZ 220-270 is used to provide century groupings for works published before 1801. == Classification categories == === Preclinical Sciences === QS Human Anatomy QT Physiology QU Biochemistry QV Pharmacology QW Microbiology & Immunology QX Parasitology QY Clinical Pathology QZ Pathology === Medicine and Related Subjects === W Health Professions WA Public Health WB Practice of Medicine WC Communicable Diseases WD Disorders of Systemic, Metabolic, or Environmental Origin, etc. WE Musculoskeletal System WF Respiratory System WG Cardiovascular System WH Hemic and Lymphatic Systems WI Digestive System WJ Urogenital System WK Endocrine System WL Nervous System WM Psychiatry WN Radiology. Diagnostic Imaging WO Surgery WP Gynecology WQ Obstetrics WR Dermatology WS Pediatrics WT Geriatrics. Chronic Disease WU Dentistry. Oral Surgery WV Otolaryngology WW Ophthalmology WX Hospitals & Other Health Facilities WY Nursing WZ History of Medicine 19th Century Schedule

  Read more →
• 

  Sense Networks

  Sense Networks is a New York City based company with a focus on applications that analyze big data from mobile phones, carrier networks, and taxicabs, particularly by using machine learning technology to make sense of large amounts of location (latitude/longitude) data. In 2009, Sense was named one of "The 25 Most Intriguing Startups in the World" by Bloomberg Businessweek and was called "The Next Google" on the cover of Newsweek. In 2014, Sense Networks was acquired by YP, "the local search and advertising company owned by Cerberus Capital Management and AT&T." It was subsequently sold off to Verve in 2017 == History == Sense Networks was founded by Greg Skibiski in February 2006 (2003?) near his home in Northampton, Massachusetts. After establishing an office in NoHo, New York City near Silicon Alley, Skibiski recruited Alex Pentland, Director of Human Dynamics Research and former Academic Head of the MIT Media Lab, Tony Jebara, Associate Professor and Head of the Machine Learning Laboratory at Columbia University, and Christine Lemke, who would later become co-founders. Sense Networks investors include Intel Capital, Javelin Venture Partners, and Kenan Altunis. Founder Greg Skibiski was pushed out by lead investor Intel Capital in November 2009 following the company's B round of financing. During the same week, the company won the Emerging Communications Conference "Company to Watch" Award. The company has three published patent applications for analyzing sensor data streams: System and Method of Performing Location Analytics (US 20090307263), Comparing Spatial-Temporal Trails in Location Analytics (US 20100079336), and Anomaly Detection in Sensor Analytics (US 20100082301). The company was acquired by the Yellow Pages in 2014. This is a marketing conglomerate under AT&T and Cerberus Capital Management. == Products and services == The Citysense consumer application that shows hotspots of human activity in real-time from mobile phone location and taxicab GPS data was named by ReadWriteWeb (in The New York Times) as "Top 10 Internet of Things Products of 2009". The Cabsense consumer application that shows the best place to catch a New York City taxicab based on GPS data from the vehicle was launched in March 2010. The Macrosense platform is for mobile application providers and mobile phone carriers to analyze billions of customer location data points for predictive analytics in advertising and churn management applications. == Privacy and data ownership == The company allows users to opt-out of their service through their website, and users may monitor their profile through their application. The company does not collect identifiable data (such as phone numbers or names); it collects data received from cellphone to construct anonymous profiles of consumers. This anonymous data/profiles may then be sold to third parties. The company's privacy and data ownership policies are based on The New Deal on Data, as advocated by Alex "Sandy" Pentland, head of the Human Dynamics group at MIT.

  Read more →
• 

  Social History and Industrial Classification

  Social History and Industrial Classification (SHIC) is a classification system used by many British museums for social history and industrial collections. It was first published in 1983. == Purpose == SHIC classifies materials (books, objects, recordings etc.) by their interaction with the people who used them. For example, a carpenter's hammer is classified with other tools of the carpenter, and not with a blacksmith's hammer. In contrast other classification systems, for example the Dewey Decimal Classification, might class all hammers together and close to the classification for other percussive tools. The specialist subject network, Social History Curator's Group (SHCG), obtained funding in 2012 to develop an on-line version, now on their website http://www.shcg.org.uk/ == Scheme == Materials are classified under four major category numbers: Community life Domestic and family life Personal life Working life Further classification within a category is by the use of further numbers after the decimal point. It is permissible to assign more than one classification in cases where the object had more than one use.

  Read more →
• 

  List of .NET libraries and frameworks

  This article contains a list of libraries that can be used in .NET languages. These languages require .NET Framework, Mono, or .NET, which provide a basis for software development, platform independence, language interoperability and extensive framework libraries. Standard Libraries (including the Base Class Library) are not included in this article. == Introduction == Apps created with .NET Framework or .NET run in a software environment known as the Common Language Runtime (CLR), an application virtual machine that provides services such as security, memory management, and exception handling. The framework includes a large class library called Framework Class Library (FCL). Thanks to the hosting virtual machine, different languages that are compliant with the .NET Common Language Infrastructure (CLI) can operate on the same kind of data structures. These languages can therefore use the FCL and other .NET libraries that are also written in one of the CLI compliant languages. When the source code of such languages are compiled, the compiler generates platform-independent code in the Common Intermediate Language (CIL, also referred to as bytecode), which is stored in CLI assemblies. When a .NET app runs, the just-in-time compiler (JIT) turns the CIL code into platform-specific machine code. To improve performance, .NET Framework also comes with the Native Image Generator (NGEN), which performs ahead-of-time compilation to machine code. This architecture provides language interoperability. Each language can use code written in other languages. Calls from one language to another are exactly the same as would be within a single programming language. If a library is written in one CLI language, it can be used in other CLI languages. Moreover, apps that consist only of pure .NET assemblies, can be transferred to any platform that contains an implementation of CLI and run on that platform. For example, apps written using .NET can run on Windows, macOS, and various versions of Linux. .NET apps or their libraries, however, may depend on native platform features, e.g. COM. As such, platform independence of .NET apps depends on the ability to transfer necessary native libraries to target platforms. In 2019, the Windows Forms and Windows Presentation Foundation portions of .NET Framework were made open source. === .NET implementations === There are four primary .NET implementations that are actively developed and maintained: .NET Framework: The original .NET implementation that has existed since 2002. While not yet discontinued, Microsoft does not plan on releasing its next major version, 5.0. Mono: A cross-platform implementation of .NET Framework by Ximian, introduced in 2004. It is free and open-source. It is now developed by Xamarin, a subsidiary of Microsoft. Universal Windows Platform (UWP): An implementation of .NET used for building UWP apps. It's designed to unify development for different targeted types of devices, including PCs, tablets, phablets, phones, and the Xbox. .NET: A cross-platform re-implementation of .NET Framework, introduced in 2016 and initially called .NET Core. It is free and open-source. .NET superseded .NET Framework with the release of .NET 5. Each implementation of .NET includes the following components: One or more runtime environments, e.g. Common Language Runtime (CLR) for .NET Framework and CoreCLR for .NET A class library The .NET Standard is a set of common APIs that are implemented in the Base Class Library of any .NET implementation. The class library of each implementation must implement the .NET Standard, but may also implement additional APIs. Traditionally, .NET apps targeted a certain version of a .NET implementation, e.g. .NET Framework 4.6. Starting with the .NET Standard, an app can target a version of the .NET Standard and then it could be used (without recompiling) by any implementation that supports that level of the standard. This enables portability across different .NET implementations. The following table lists the .NET implementations that adhere to the .NET Standard and the version number at which each implementation became compliant with a given version of .NET Standard. For example, according to this table, .NET Core 3.0 was the first version of .NET Core that adhered to .NET Standard 2.1. This means that any version of .NET Core bigger than 3.0 (e.g. .NET Core 3.1) also adheres to .NET Standard 2.1. == Web frameworks == === ASP.NET === First released in 2002, ASP.NET is an open-source server-side web application framework designed for web development to produce dynamic web pages. It is the successor to Microsoft's Active Server Pages (ASP) technology, built on the Common Language Runtime (CLR). === ASP.NET Core === ASP.NET was completely rewritten in 2016 as a modular web framework, together with other frameworks like Entity Framework. The re-written framework uses the new open-source .NET Compiler Platform (also known by its codename "Roslyn") and is cross platform. The programming models ASP.NET MVC, ASP.NET Web API, and ASP.NET Web Pages (a model using only Razor pages) were merged into a unified MVC 6. === Blazor === Blazor is a free and open-source web framework that enables developers to create Single-page Web apps using C# and HTML in ASP.NET Razor pages ("components"). Blazor is part of the ASP.NET Core framework. Blazor Server apps are hosted on a web server, while Blazor WebAssembly apps are downloaded to the client's web browser before running. In addition, a Blazor Hybrid framework is available with server-based and client-based application components. == Numerical libraries == === Open-source numerical libraries === ==== AForge.NET ==== This is a computer vision and artificial intelligence library. It implements a number of genetic, fuzzy logic and machine learning algorithms with several architectures of artificial neural networks with corresponding training algorithms. ==== ALGLIB ==== This is a cross-platform open source numerical analysis and data processing library. It consists of algorithm collections written in different programming languages (C++, C#, FreePascal, Delphi, VBA) and has dual licensing – commercial and GPL. ==== Math.NET Numerics ==== This library aims to provide methods and algorithms for numerical computations in science, engineering and everyday use. Covered topics include special functions, linear algebra, probability models, random numbers, interpolation, integral transforms and more. MIT/X11 license. ==== Meta.Numerics ==== This is a library for advanced scientific computation in the .NET Framework. ==== ML.NET ==== This is a free software machine learning library. The preview release of ML.NET included transforms for feature engineering like n-gram creation, and learners to handle binary classification, multi-class classification, and regression tasks. Additional ML tasks like anomaly detection and recommendation systems have since been added, and other approaches like deep learning will be included in future versions. === Proprietary numerical libraries === ==== ILNumerics.Net ==== This is a high performance, typesafe numerical array set of classes and functions for general math, FFT and linear algebra. The library, developed for .NET/Mono, aims to provide 32- and 64-bit script-like syntax in C#, 2D & 3D plot controls, and efficient memory management. It is released under GPLv3 or commercial license. ==== Measurement Studio ==== This is an integrated suite of UI controls and class libraries for use in developing test and measurement applications. The analysis class libraries provide various digital signal processing, signal filtering, signal generation, peak detection, and other general mathematical functionality. ==== NMath ==== This is a numerical component library for the .NET platform developed by CenterSpace Software. It includes signal processing (FFT) classes, a linear algebra (LAPACK & BLAS) framework, and a statistics package. == 3D graphics == === Open-source 3D graphics === ==== Open Toolkit (OpenTK) ==== This is a low-level C# binding for OpenGL, OpenGL ES and OpenAL. It runs on Windows, Linux, Mac OS X, BSD, Android and iOS. It can be used standalone or integrated into a GUI. ==== Windows Presentation Foundation (WPF) ==== This is a graphical subsystem for rendering user interfaces, developed by Microsoft. It also contains a 3D rendering engine. In addition, interactive 2D content can be overlaid on 3D surfaces natively. It only runs on Windows operating systems. === Proprietary 3D graphics === ==== Unity ==== This is a cross-platform game engine developed by Unity Technologies and used to develop video games for PC, consoles, mobile devices and websites. == Image processing == === AForge.NET === This is a computer vision and artificial intelligence library. It implements a number of image processing algorithms and filters. It is released under the LGPLv3 and partly GPLv3 license. Majority of the library is written in C# and th

  Read more →
• 

  Inbenta

  Inbenta is an AI company that originated in Barcelona, Spain, in 2005. Inbenta is currently headquartered in Allen, Texas, with additional offices in Spain, São Paulo, Brazil, Toulouse, France, and Tokyo, Japan. Inbenta provides natural language processing and semantic search through artificial intelligence. == History == Inbenta raised $12 Million in their Series B funding round to extend the reach of their artificial intelligence for business solutions. In 2023 Inbenta's new chief executive officer Melissa Solis moved Inbenta's headquarters to One Bethany West in Allen, Texas from Foster City, California. == Controversy == On 23 June 2018, Ticketmaster UK identified malicious software on a customer support product hosted by Inbenta Technologies, compromising personal data and payment details for thousands of Ticketmaster customers. Three days later, Inbenta's CEO Issued a message about the incident to convey the full scope of the breach. Also on its FAQ section, Inbenta claimed that "After a careful analysis of all clues and snapshots from our systems, the technical team at Inbenta discovered that the script had been implemented on the payment page. We were unaware of this, and would have advised against doing so had we known, as it presents a point of vulnerability". On November 13, 2020, the Information Commissioner's Office fined Ticketmaster UK Limited £1.25 million for failing to protect customers' payment details. According to the ICO, "It was because of Ticketmaster's business decision to include the [Inbenta] chat bot on its payment page that the chat bot was able to unlawfully process the personal data of customers."

  Read more →
• 

  Catastrophic interference

  Catastrophic interference, also known as catastrophic forgetting, is the tendency of an artificial neural network to abruptly and drastically forget previously learned information upon learning new information. Neural networks are an important part of the connectionist approach to cognitive science. The issue of catastrophic interference when modeling human memory with connectionist models was originally brought to the attention of the scientific community by research from McCloskey and Cohen (1989), and Ratcliff (1990). It is a radical manifestation of the 'sensitivity-stability' dilemma or the 'stability-plasticity' dilemma. Specifically, these problems refer to the challenge of making an artificial neural network that is sensitive to, but not disrupted by, new information. Lookup tables and connectionist networks lie on the opposite sides of the stability plasticity spectrum. The former remains completely stable in the presence of new information but lacks the ability to generalize, i.e. infer general principles, from new inputs. On the other hand, connectionist networks like the standard backpropagation network can generalize to unseen inputs, but they are sensitive to new information. Backpropagation models can be analogized to human memory insofar as they have a similar ability to generalize, but these networks often exhibit less stability than human memory. Notably, these backpropagation networks are susceptible to catastrophic interference. This is an issue when modelling human memory, because unlike these networks, humans typically do not show catastrophic forgetting. == Discovery == The term catastrophic interference was originally coined by McCloskey and Cohen (1989) but was also brought to the attention of the scientific community by research from Ratcliff (1990). === The Sequential Learning Problem: McCloskey and Cohen (1989) === McCloskey and Cohen (1989) noted the problem of catastrophic interference during two different experiments with backpropagation neural network modelling. Experiment 1: Learning the ones and twos addition facts In their first experiment they trained a standard backpropagation neural network on a single training set consisting of 17 single-digit ones problems (i.e., 1 + 1 through 9 + 1, and 1 + 2 through 1 + 9) until the network could represent and respond properly to all of them. The error between the actual output and the desired output steadily declined across training sessions, which reflected that the network learned to represent the target outputs better across trials. Next, they trained the network on a single training set consisting of 17 single-digit twos problems (i.e., 2 + 1 through 2 + 9, and 1 + 2 through 9 + 2) until the network could represent, respond properly to all of them. They noted that their procedure was similar to how a child would learn their addition facts. Following each learning trial on the twos facts, the network was tested for its knowledge on both the ones and twos addition facts. Like the ones facts, the twos facts were readily learned by the network. However, McCloskey and Cohen noted the network was no longer able to properly answer the ones addition problems even after one learning trial of the twos addition problems. The output pattern produced in response to the ones facts often resembled an output pattern for an incorrect number more closely than the output pattern for a correct number. This is considered to be a drastic amount of error. Furthermore, the problems 2+1 and 1+2, which were included in both training sets, even showed dramatic disruption during the first learning trials of the twos facts. Experiment 2: Replication of Barnes and Underwood (1959) study In their second connectionist model, McCloskey and Cohen attempted to replicate the study on retroactive interference in humans by Barnes and Underwood (1959). They trained the model on A-B and A-C lists and used a context pattern in the input vector (input pattern), to differentiate between the lists. Specifically the network was trained to respond with the right B response when shown the A stimulus and A-B context pattern and to respond with the correct C response when shown the A stimulus and the A-C context pattern. When the model was trained concurrently on the A-B and A-C items then the network readily learned all of the associations correctly. In sequential training the A-B list was trained first, followed by the A-C list. After each presentation of the A-C list, performance was measured for both the A-B and A-C lists. They found that the amount of training on the A-C list in Barnes and Underwood study that lead to 50% correct responses, lead to nearly 0% correct responses by the backpropagation network. Furthermore, they found that the network tended to show responses that looked like the C response pattern when the network was prompted to give the B response pattern. This indicated that the A-C list apparently had overwritten the A-B list. This could be likened to learning the word dog, followed by learning the word stool and then finding that you think of the word stool when presented with the word dog. McCloskey and Cohen tried to reduce interference through a number of manipulations including changing the number of hidden units, changing the value of the learning rate parameter, overtraining on the A-B list, freezing certain connection weights, changing target values 0 and 1 instead 0.1 and 0.9. However, none of these manipulations satisfactorily reduced the catastrophic interference exhibited by the networks. Overall, McCloskey and Cohen (1989) concluded that: at least some interference will occur whenever new learning alters the weights involved in representing old learning the greater the amount of new learning, the greater the disruption in old knowledge interference was catastrophic in the backpropagation networks when learning was sequential but not concurrent === Constraints Imposed by Learning and Forgetting Functions: Ratcliff (1990) === Ratcliff (1990) used multiple sets of backpropagation models applied to standard recognition memory procedures, in which the items were sequentially learned. After inspecting the recognition performance models he found two major problems: Well-learned information was catastrophically forgotten as new information was learned in both small and large backpropagation networks. Even one learning trial with new information resulted in a significant loss of the old information, paralleling the findings of McCloskey and Cohen (1989). Ratcliff also found that the resulting outputs were often a blend of the previous input and the new input. In larger networks, items learned in groups (e.g. AB then CD) were more resistant to forgetting than were items learned singly (e.g. A then B then C...). However, the forgetting for items learned in groups was still large. Adding new hidden units to the network did not reduce interference. Discrimination between the studied items and previously unseen items decreased as the network learned more. This finding contradicts studies on human memory, which indicated that discrimination increases with learning. Ratcliff attempted to alleviate this problem by adding 'response nodes' that would selectively respond to old and new inputs. However, this method did not work as these response nodes would become active for all inputs. A model which used a context pattern also failed to increase discrimination between new and old items. == Proposed solutions == The main cause of catastrophic interference seems to be overlap in the representations at the hidden layer of distributed neural networks. In a distributed representation, each input tends to create changes in the weights of many of the nodes. Catastrophic forgetting occurs because when many of the weights where "knowledge is stored" are changed, it is unlikely for prior knowledge to be kept intact. During sequential learning, the inputs become mixed, with the new inputs being superimposed on top of the old ones. Another way to conceptualize this is by visualizing learning as a movement through a weight space. This weight space can be likened to a spatial representation of all of the possible combinations of weights that the network could possess. When a network first learns to represent a set of patterns, it finds a point in the weight space that allows it to recognize all of those patterns. However, when the network then learns a new set of patterns, it will move to a place in the weight space for which the only concern is the recognition of the new patterns. To recognize both sets of patterns, the network must find a place in the weight space suitable for recognizing both the new and the old patterns. Below are a number of techniques which have empirical support in successfully reducing catastrophic interference in backpropagation neural networks: === Orthogonality === Many of the early techniques in reducing representational overlap involved making either the input vecto

  Read more →