AI For Psychology Students

AI For Psychology Students — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

Supper (Spotify)

Supper is a web-based application on the Spotify digital music streaming platform. The Supper app was born from a group of friends who had backgrounds in the music and gastronomy industries. Digital music solutions company Artisan Council later executed it. The app now sits in the top 40 applications on Spotify. == About == The Supper Spotify application matches recipes for all occasions and skill levels with a playlist for both preparation and presentation, as envisioned by the chefs themselves. Supper is credited with being one of the first apps to pair music with food. Playing on the social nature of music and food culture, users can seamlessly experience both for the first time with real time music streaming. == Supper.mx == In May 2014 Supper was launched outside of the Spotify streaming platform. Though still in partnership with Spotify, supper.mx allows users to view Supper's music + food collaborations on mobile, tablet and desktop, without the need to download Spotify directly. == Curators == All of the recipes and playlists featured on the Supper app come straight from a growing network of tastemakers, including chefs, musicians and institutions around the world. Each month the recipes and playlists are updated in conjunction with current holidays, events and seasons. === Launch === Launching in October 2013 the first edition of Supper featured content from a range of eating institutions and culture makers from the US and Australia. Brooklyn Bowl (Brooklyn) Roberta's Pizza (Brooklyn) Fancy Hanks (Melbourne) The Foresters/Queenies Upstairs (Sydney) Hipstamatic Panama House (Bondi) Sweetwater Inn (Melbourne) Soul Clap (Syd record label) Yellow Birds (Melbourne) === November 2013 === Yardbird (Hong Kong) Sonoma Bakery (Sydney) Do or Dine (Brooklyn) Cameo Gallery (Brooklyn) Hypertrak (Blog) Blue Smoke (NYC) The Crepes of Wrath (Blog) Willin Low // Wild Rocket - Wild Oats - Relish === December 2013 === The Copper Mill (Sydney) Thug Kitchen Mamak (Sydney) Tutu's (Brooklyn) Chin Chin (Melbourne) Flat Iron Steak (London) Greasy Spoon (Copenhagen) === January 2014 === Mexicali Taco & Co. (LA) Church & State (LA) Salts Cure (LA) Nopa (SF) L & E Oyster (LA) 4100 bar (LA) Golden Gopher (LA) The Pie Hole (LA) State Bird Provisions (SF) === Momofuku === In February 2014 Supper teamed up with restaurant heavy weights Momofuku. The recipes featured came from their iconic New York, Toronto and Sydney restaurants. Head office also got involved with an instructional from Brand Director Sue Chan on how to paint Momofuku vibes on to any party. === SXSW === March sees the Supper team migrate to Austin, Texas for SXSW, bringing together the best eateries the city has to offer as well as the music that has influenced them. Restaurants and eateries on board in 2014 included: The Backspace Kelis Swifts Attic Uchi Jackalope Paul Qui/East Side King Thai Kun Wonderland Hole in the Wall Justine's Brasserie The Liberty === Kelis === In April 2014 Kelis presented 5 of her recipes paired with a personal playlist for Supper. Kelis shared her recipes for apple farro, jerk ribs, New York vanilla bean cheesecake and Jerk Ribs. The Kelis/Supper collaboration coincided with the release of Kelis' 2014 album titled 'Food'. === Roberta's Pizza === In May 2014 Bushwick's Roberta's Pizza was guest curator on the Supper app and website. Included in their selections were restaurants and bars from across New York including Bun-ker Vietnamese, Old Stanley's Bar, St. Anselm, Chuko, Frank's Cocktail Lounge, Junior's Cheesecake, Xi'an Famous Foods, Xe Lua, 124 Old Rabbit and Yuji Ramen.
Read more →
Kindwise

FlowerChecker, also known as Kindwise, is a company that uses machine learning to identify natural objects from images. This includes plants and their diseases, but also insects and mushrooms. It is based in Brno, Czech Republic. It was founded in 2014 by Ondřej Veselý, Jiří Řihák, and Ondřej Vild, at the time Ph.D. students. == Features & Tools == FlowerChecker offers multiple products. Plant.id is a machine learning-based plant identification API launched in 2018, with the plant disease identification API, plant.health, released in April 2022. The plant.id API is suitable for integration into other software, such as mobile apps or urban trees from remote-sensing imagery. Other products include insect.id, mushroom.id and crop.health are machine learning-based identification APIs for the identification of insects, fungi and economically important plants, respectively, and include also online public demos. The FlowerChecker app was discontinued in October 2024 after 10 years of successful operation. == Recognition == In 2019, FlowerChecker won the Idea of the Year award in the AI Awards organized by the Confederation of Industry of the Czech Republic. In 2020, an academic study comparing ten free automated image recognition apps showed that plant.id's performance excelled in most of the parameters studied. In an independent study comparing different image-based species recognition models and their suitability for recognizing invasive alien species, the plant.id achieved the highest accuracy compared to other tools. In a subsequent study, plant.id was utilized to evaluate urban forest biodiversity using remote-sensing imagery, achieving the highest accuracy in tree species identification among compared methods. The technology has also been referenced as an example of practical integration of AI-based plant identification into cross-platform precision agriculture systems. == Research activities == Flowerchecker cooperates with the Nature Conservation Agency of the Czech Republic on a biodiversity mapping project. FlowerChecker plans to adapt its services to participate in the control of invasive species. In 2022, the company entered a consortium to develop a weeder capable of in-row weed detection and removal. In 2025, it received funding for the development of a technology for the removal of invasive species.
Read more →
Is-a

In knowledge representation, ontology components and ontology engineering, including for object-oriented programming and design, is-a (also written as is_a or is a) is a subsumptive relationship between abstractions (e.g., types, classes), wherein one class A is a subclass of another class B (and so B is a superclass of A). In other words, type A is a subtype of type B when A's specification implies B's specification. That is, any object (or class) that satisfies A's specification also satisfies B's specification, because B's specification is weaker. For example, a cat 'is a[n]' animal, but not vice versa. All cats are animals, but not all animals are cats. Behaviour that is relevant to all animals is defined on an animal class, whereas behaviour that is relevant only for cats is defined in a cat class. By defining the cat class as 'extending' the animal class, all cats 'inherit' the behaviour defined for animals, without the need to explicitly code that behaviour for cats. == Related concepts == The is-a relationship is to be contrasted with the has-a (has_a or has a) relationship between types (classes); confusing the relations has-a and is-a is a common error when designing a model (e.g., a computer program) of the real-world relationship between an object and its subordinate. The is-a relationship may also be contrasted with the instance-of relationship between objects (instances) and types (classes): see Type–token distinction. To summarize the relations, there are: hyperonym–hyponym (supertype/superclass–subtype/subclass) relations between types (classes) defining a taxonomic hierarchy, where for a subsumption relation: a hyponym (subtype, subclass) has a type-of (is-a) relationship with its hyperonym (supertype, superclass); holonym–meronym (whole/entity/container–part/constituent/member) relations between types (classes) defining a possessive hierarchy, where for an aggregation (i.e. without ownership) relation: a holonym (whole) has a has-a relationship with its meronym (part), for a composition (i.e. with ownership) relation: a meronym (constituent) has a part-of relationship with its holonym (entity), for a containment relation: a meronym (member) has a member-of relationship with its holonym (container); concept–object (type–token) relations between types (classes) and objects (instances), where a token (object) has an instance-of relationship with its type (class).
Read more →
Attribute–value system

An attribute–value system is a basic knowledge representation framework comprising a table with columns designating "attributes" (also known as "properties", "predicates", "features", "dimensions", "characteristics", "fields", "headers" or "independent variables" depending on the context) and "rows" designating "objects" (also known as "entities", "instances", "exemplars", "elements", "records" or "dependent variables"). Each table cell therefore designates the value (also known as "state") of a particular attribute of a particular object. == Example of attribute–value system == Below is a sample attribute–value system. It represents 10 objects (rows) and five features (columns). In this example, the table contains only integer values. In general, an attribute–value system may contain any kind of data, numeric or otherwise. An attribute–value system is distinguished from a simple "feature list" representation in that each feature in an attribute–value system may possess a range of values (e.g., feature P1 below, which has domain of {0,1,2}), rather than simply being present or absent (Barsalou & Hale 1993). == Other terms used for "attribute–value system" == Attribute–value systems are pervasive throughout many different literatures, and have been discussed under many different names: Flat data Spreadsheet Attribute–value system (Ziarko & Shan 1996) Information system (Pawlak 1981) Classification system (Ziarko 1998) Knowledge representation system (Wong & Ziarko 1986) Information table (Yao & Yao 2002)
Read more →
Outline of deep learning

The following outline is provided as an overview of, and topical guide to, deep learning: Deep learning is a subfield of machine learning and artificial intelligence based on artificial neural networks with multiple processing layers. It emphasizes representation learning and is widely used in areas such as computer vision, natural language processing, speech recognition, recommender systems, robotics, and generative artificial intelligence. == Ways to categorize deep learning == A field of study A branch of artificial intelligence A subfield of machine learning A subfield of computer science A form of representation learning A class of methods based on artificial neural networks An approach used in computational statistics == History == === Precursors === Cybernetics Perceptron Connectionism Neocognitron Backpropagation === Milestones === LeNet Long short-term memory Deep belief network AlexNet Sequence to sequence learning Generative adversarial network Residual neural network Transformer BERT Generative pre-trained transformer Diffusion model === Related histories === History of artificial intelligence History of machine learning Timeline of machine learning == Core concepts == == Learning settings == Supervised learning Unsupervised learning Self-supervised learning Semi-supervised learning Reinforcement learning Transfer learning Multitask learning Multimodal learning Online machine learning Continual learning == Common tasks == Image classification Object detection Image segmentation Automatic speech recognition Neural machine translation Question answering Automatic summarization Text-to-image model Protein structure prediction == Architectures == === Feedforward and convolutional architectures === Feedforward neural network Multilayer perceptron Convolutional neural network Radial basis function network Residual neural network U-Net === Recurrent and sequence architectures === Recurrent neural network Long short-term memory Gated recurrent unit Sequence to sequence learning Recursive neural network === Representation-learning architectures === Autoencoder Denoising autoencoder Sparse autoencoder Variational autoencoder Restricted Boltzmann machine Deep belief network === Attention and transformer architectures === Attention (machine learning) Transformer BERT Generative pre-trained transformer Vision transformer === Generative and probabilistic architectures === Autoregressive model Diffusion model Energy-based model Generative adversarial network Mixture of experts === Graph and memory architectures === Graph neural network Graph convolutional network Siamese network Neural Turing machine Memory network Echo state network Capsule neural network == Neural network components and techniques == Artificial neuron Activation function Rectified linear unit Sigmoid function Softmax function Embedding Convolution Pooling layer Attention Batch normalization Layer normalization Residual connections == Training and optimization == Backpropagation Gradient descent Stochastic gradient descent Adam optimization Learning rate Loss function Cross-entropy Mean squared error Regularization Dropout Early stopping Batch normalization Data augmentation Transfer learning Knowledge distillation Ensemble learning Curriculum learning == Datasets and benchmarks == CIFAR-10 ImageNet MNIST database Common Objects in Context (COCO) General Language Understanding Evaluation (GLUE) benchmark LibriSpeech SQuAD == Applications == === Computer vision === Computer vision Facial recognition system Image classification Image segmentation Medical imaging Object detection Optical character recognition === Natural language processing === Automatic summarization Chatbot Information retrieval Large language model Natural language processing Neural machine translation Question answering Sentiment analysis === Speech and audio === Automatic speech recognition Music information retrieval Speaker recognition Speech synthesis === Science and medicine === Bioinformatics Computational biology Drug discovery Medical diagnosis Protein structure prediction === Robotics and control === Autonomous car Computer game bot Control theory Robotics === Recommendation, search, and forecasting === Anomaly detection Forecasting Fraud detection Recommender system Search engine === Generative artificial intelligence === Deepfake Generative artificial intelligence Large language model Speech synthesis Text-to-image model === Computer graphics and video games === Deep Learning Anti-Aliasing (DLAA) Deep Learning Super Sampling (DLSS) == Hardware == AMD Instinct AMD XDNA Application-specific integrated circuit Deep learning processor, Neural processing unit (NPU), or Neural Engine Field-programmable gate array General-purpose computing on graphics processing units (GPGPU) Graphics processing unit NVIDIA Deep Learning Accelerator (NVDLA) Tensor processing unit Vision processing unit Wafer-scale integration === Supporting software platforms === CUDA Metal ROCm == Software == === Open-source frameworks and libraries === === Neural network software === EDLUT Emergent Encog JOONE Neuroph NeuroSolutions OpenNN Peltarion Synapse SNNS === Platforms, tools, and deployment === Amazon SageMaker Google Colab Hugging Face Kaggle Kubeflow MLflow ONNX OpenVINO TensorFlow Hub == Algorithms for deep learning and neural networks == Backpropagation Conjugate gradient method Generalized Hebbian algorithm Gradient descent Levenberg–Marquardt algorithm Perceptron Quasi-Newton method Wake-sleep algorithm == Methods and related topics == === Representation and metric learning === Contrastive learning Embedding Feature learning Manifold learning Metric learning === Generative modeling === Autoregressive model Diffusion model Generative adversarial network Generative model Variational inference === Efficient and scalable deep learning === Knowledge distillation Low-rank approximation Mixture of experts Quantization Sparsity === Reliability, safety, and interpretability === Adversarial machine learning AI alignment Algorithmic bias Catastrophic forgetting Differential privacy Explainable artificial intelligence Federated learning Hallucination (artificial intelligence) == Conferences and workshops == Annual Meeting of the Association for Computational Linguistics Conference on Computer Vision and Pattern Recognition Conference on Neural Information Processing Systems International Conference on Computer Vision International Conference on Learning Representations International Conference on Machine Learning == Organizations == === Research laboratories and institutions === Allen Institute for AI Alberta Machine Intelligence Institute European Laboratory for Learning and Intelligent Systems Google DeepMind Meta AI Mila Microsoft Research Vector Institute === Companies === Anthropic Cerebras Cohere DeepSeek Mistral AI OpenAI Stability AI xAI == Publications == === Books === Deep Learning – Ian Goodfellow and Yoshua Bengio Neural Networks and Deep Learning – Michael Nielsen Perceptrons – Marvin Minsky and Seymour Papert === Journals === IEEE Transactions on Neural Networks and Learning Systems Neural Networks Neural Computation == Influential persons ==
Read more →
David Krueger (professor)

David Krueger is an American machine learning professor and advocate for the reduction of risks related to artificial intelligence. Krueger is an assistant professor in Robust, Reasoning, and Responsible AI at the University of Montreal and a Core Academic Member at Mila. == Early life and education == Krueger obtained a B.A. in mathematics from Reed College, and completed his MSc and Ph.D. in Computer Science at the University of Montreal. He trained in deep learning under Yoshua Bengio, Roland Memisevic, and Aaron Courville from 2013 to 2021. Krueger was also an intern on Google DeepMind's AI Safety team in 2018. == Career == Krueger researches deep learning, AI alignment, and AI safety. His work is focused on reducing the risk of human extinction resulting from out-of-control AI systems. Krueger was an assistant professor at the University of Cambridge from 2021 to 2024, before taking a faculty position at the University of Montreal in 2024. In 2023, he was a founding research director at the UK AI Security Institute. That same year, Krueger initiated the Statement on AI Risk, which argues that AI could cause human extinction and was signed by Anthropic's Dario Amodei, OpenAI's Sam Altman, AI expert Geoffrey Hinton, and other leaders. In April 2026, Krueger discussed the risks of advanced AI at a Capitol Hill event hosted by Senator Bernie Sanders. === Evitable === In 2025, Krueger founded Evitable, a nonprofit organization that advocates for an AI moratorium. == Views == Krueger argues that AI will lead to a "gradual disempowerment" of workers, likening AI chips to nuclear bombs. He also says the military use of AI "poses an existential risk to humanity."
Read more →
TRAIGA

TRAIGA, or the Texas Responsible Artificial Intelligence Governance Act, is a state law regulating the development and deployment of artificial intelligence (AI) systems in Texas. Sponsored by Representative Giovanni Capriglione, the Act establishes a framework governing certain uses of AI, outlines prohibited uses, and creates obligations on state government entities, among other provisions. TRAIGA was signed into law in 2025 and took effect on January 1, 2026. The law applies to AI developers and deployers that conduct business in Texas or whose systems are used by Texas residents. It prohibits the intentional development or deployment of AI systems to incite harm, violate constitutional rights, engage in unlawful discrimination, and produce child sexual abuse material or unlawful deepfakes. TRAIGA also establishes the Texas Artificial Intelligence Council and creates a regulatory sandbox program. The Texas Attorney General is charged with enforcement. It has received attention as one of the first comprehensive AI-related laws enacted by a U.S. state. Legal analysts have compared it to the European Union (EU) Artificial Intelligence Act and the Colorado AI Act, noting its intent-based discrimination standard and narrower scope relative to those frameworks. == Background == In June 2023, Texas Governor Greg Abbott signed House Bill 2060, which created an Artificial Intelligence Advisory Council within the Texas Department of Information Resources. The Council was tasked with monitoring the use of AI systems across state government. Its membership included representatives from law enforcement, academia, and the legal profession. After submitting a report to state policymakers, the Council was disbanded in December 2024. Separately, the Texas House Select Committee on Artificial Intelligence and Emerging Technologies was created in 2023 to examine the political and social implications of artificial intelligence. Among its recommendations was the creation of a regulatory sandbox to allow for controlled testing of AI systems. This recommendation informed the regulatory sandbox provision included in TRAIGA. == History == In December 2024, Representative Capriglione introduced House Bill 1709, the Texas Responsible Artificial Intelligence Governance Act. The bill sought to create a statewide framework for artificial intelligence, including transparency requirements for companies deploying AI systems, restrictions on certain uses of AI, and the creation of a regulatory sandbox. Modeled in part on the EU Artificial Intelligence Act and the Colorado AI Act, House Bill 1709 focused on "high-risk" AI systems and included provisions addressing private sector liability. House Bill 1709 did not advance during the legislative session. Industry stakeholders raised concerns that several provisions were overly burdensome. The bill informed the development of a revised proposal, House Bill 149, also titled the Texas Responsible Artificial Intelligence Governance Act. The revised version removed requirements for private companies to notify consumers when they interact with AI systems and to conduct impact assessments, among other provisions. In April 2025, an amended version of House Bill 149 passed the Texas House of Representatives and was referred to the Senate Committee on Business and Commerce. The bill later received approval from both chambers, where the House voted on amendments adopted by the Senate. On May 31, 2025, the state legislature passed House Bill 149, one of several AI-related bills considered during the legislative session. Governor Abbott signed TRAIGA into law on June 22, 2025. During the legislative process, a proposed federal moratorium on state-level AI regulation initially raised questions about the enforceability of state AI laws, including TRAIGA. At the time of signing, Governor Abbott stated that Texas would ensure compliance with applicable federal requirements. In July 2025, the United States Senate voted to remove the proposed moratorium from federal legislation. The Act took effect on January 1, 2026. == Provisions == === Definitions and scope === TRAIGA applies to AI developers and deployers that advertise or conduct business in Texas, develop products used by Texas residents, or develop or deploy AI systems within the state. The Act also applies to Texas state and local government entities. The Act defines a developer as a person who develops an AI system and a deployer as one who deploys an AI system in Texas. Consumers are defined as Texas residents. The Act defines an artificial intelligence system as a machine-based system that "infers from the inputs the system receives how to generate outputs, including content, decisions, predictions, or recommendations, that can influence physical or virtual environments." === Government use === The Act requires government agencies to provide consumers with plain language notices before interacting with AI systems. It also prohibits government agencies from using artificial intelligence systems to assign social scores to consumers. It also restricts the use of AI systems to identify individuals using biometric data without the individual’s consent. === Prohibitions === The Act prohibits the development or deployment of artificial intelligence systems intended to cause harm, self-harm, or criminal activity. It also prohibits the development or deployment of AI systems designed to violate constitutional rights or unlawfully discriminate based on protected classes. In addition, the Act prohibits the development or deployment of AI systems that are intended to produce or distribute child sexual abuse material or unlawful deepfakes. === Enforcement === Enforcement authority under the Act rests with the Texas Attorney General. The Act does not create a private right of action. The Act requires the Texas Attorney General to create an online complaint system where consumers may submit allegations of potential violations. The Attorney General can investigate complaints received through this system and may request information relevant to the operation of an AI system, including information about training data. Before initiating an enforcement action, the Attorney General must provide a written notice to the alleged violator, who is then provided with a 60-day period to cure the alleged violation. === Penalties === If a violation is not cured, the Act authorizes civil penalties. Penalties range from $10,000 to $12,000 per curable violation and from $80,000 to $200,000 per non-curable violation. The Act also authorizes additional penalties of $2,000 to $40,000 for each day the violation continues. If the Attorney General determines that a person certified or licensed by a state agency has violated the Act and recommends enforcement, the relevant agency may impose additional administrative sanctions, including license suspension or further monetary penalties. === Safe harbor === The Act provides an affirmative defense for AI developers and deployers who identify potential violations through internal testing or auditing or who demonstrate compliance with National Institute of Standards and Technology (NIST)'s Artificial Intelligence Risk Management Framework or a comparable risk management framework. The Act also affords protection to developers and deployers when a third party uses their AI systems in a way that violates the Act. === Texas Artificial Intelligence Council === The Act creates the Texas Artificial Intelligence Council to assist the state legislatures in evaluating artificial intelligence policy and oversight. The Council is charged with developing recommendations for state agencies regarding the use of AI systems and with overseeing the regulatory sandbox. TRAIGA gives the Council the ability to organize AI-related training for state entities and issue reports concerning artificial intelligence. The Council does not have binding rulemaking authority. The Council consists of seven members appointed by the governor, the lieutenant governor, and the speaker of the Texas House of Representatives. === Regulatory sandbox === The Act directs the Texas Department of Information Resources to create a regulatory sandbox program that allows participants to test AI systems under state supervision in a modified regulatory setting. To join the program, companies must submit applications that describe their AI systems and intended use. Approved participants may operate within the sandbox for up to 36 months. During that period, the Attorney General is restricted from initiating enforcement actions for certain categories of violations. == Reception == === Support === During legislative testimony, the Texas Public Policy Foundation stated that TRAIGA would benefit Texas businesses by reducing legal ambiguity and creating clearer compliance standards. Representatives of business groups also expressed support, stating that the Act would not impose overly burdensome regulations. The consum
Read more →
Unified Modeling Language

The Unified Modeling Language (UML) is a general-purpose, object-oriented, visual modeling language that provides a way to visualize the architecture and design of a system, similar to the function of a blueprint. UML defines notation for many types of diagrams which focus on aspects such as behavior, interaction, and structure. UML is both a formal metamodel and a collection of graphical templates. The metamodel defines the elements in an object-oriented model such as classes and properties. It is essentially the same thing as the metamodel in object-oriented programming (OOP), however for OOP, the metamodel is primarily used at run time to dynamically inspect and modify an application object model. The UML metamodel provides a mathematical, formal foundation for the graphic views used in the modeling language to describe an emerging system. UML was created in an attempt to define a standard language for object-oriented programming at the OOPSLA '95 Conference. Originally, Grady Booch and James Rumbaugh merged their models into a unified model. This was followed by Booch's company Rational Software purchasing Ivar Jacobson's Objectory company and merging their model into the UML. At the time Rational and Objectory were two of the dominant players in the small world of independent vendors of object-oriented tools and methods. The Object Management Group (OMG) then took ownership of UML. The creation of UML was motivated by the desire to standardize the disparate nature of notational systems and approaches to software design at the time. In 1997, UML was adopted as a standard by the Object Management Group (OMG) and has been managed by this organization ever since. In 2005, UML was also published by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) as the ISO/IEC 19501 standard. Since then the standard has been periodically revised to cover the latest revision of UML. Most developers do not use UML per se, but instead produce more informal diagrams, often hand-drawn. These diagrams, however, often include elements from UML. == Use == UML is primarily used for software development (in any industry or domain) but also used outside elsewhere including business processes, system functions, database schemas, workflow in the legal systems, medical electronics, Health care systems, and hardware design.. The UML is used by the OMG itself to define other OMG products such as the Unified Architecture Framework (UAF) and the Systems Modelling Language (SysML) v1. UML is designed for use with many object-oriented software development methods, both today and for the methods when it was first developed – including OMT, Booch method, Objectory, and especially RUP, which it was originally intended to be used with when work began at Rational Software. Although originally intended for object-oriented design documentation, UML has been used effectively in other contexts such as modeling business process. As UML is not inherently linked to a particular programming language, it can be used for modeling a system independent of language. Some UML tools generate source code from a UML model. === Elements === UML diagrams support visualizing system aspects like: Use case diagram for specifying user interactions with systems Class diagram for specifying structures, including data structures Activity diagram for specifying business process workflows Component diagram for specifying how components interface with other components Deployment diagram for specifying how components are deployed and executed on computational nodes In addition to syntactical (notational) elements with well-defined semantics, UML diagrams also allow for free-form comments (notes) that explain aspects such as usage, constraints, and intents. === Sharing === UML models can be exchanged among UML tools via the XML Metadata Interchange (XMI) format. === Cardinality notation === As with database Chen, Bachman, and ISO ER diagrams, class models are specified to use "look-across" cardinalities, even though several authors (Merise, Elmasri & Navathe, amongst others) prefer same-side or "look-here" for roles and both minimum and maximum cardinalities. Recent researchers (Feinerer and Dullea et al.) have shown that the "look-across" technique used by UML and ER diagrams is less effective and less coherent when applied to n-ary relationships of order strictly greater than 2. Feinerer says: "Problems arise if we operate under the look-across semantics as used for UML associations. Hartmann investigates this situation and shows how and why different transformations fail.", and: "As we will see on the next few pages, the look-across interpretation introduces several difficulties which prevent the extension of simple mechanisms from binary to n-ary associations." === Artifacts === An artifact is the "specification of a physical piece of information that is used or produced by a software development process, or by deployment and operation of a system" including models, source code, scripts, executables, tables in database systems, development deliverables, a design documents, and email messages. An artifact is the physical entity that is deployed to a node. Other UML elements such as classes and components are first manifest into artifacts and instances of these artifacts are then deployed. Artifacts can be composed of other artifacts. === Metamodeling === The OMG developed a metamodeling architecture to define UML, called the Meta-Object Facility (MOF). MOF is designed as a four-layered architecture, as shown in the image at right. It provides a meta-meta model at the top, called the M3 layer. This M3-model is the language used by Meta-Object Facility to build metamodels, called M2-models. The most prominent example of a Layer 2 Meta-Object Facility model is the UML metamodel, which describes UML itself. These M2-models describe elements of the M1-layer, and thus M1-models. These would be, for example, models written in UML. The last layer is the M0-layer or data layer. It is used to describe runtime instances of the system. The metamodel can be extended using a mechanism called stereotyping. This has been criticized as being insufficient/untenable by Brian Henderson-Sellers and Cesar Gonzalez-Perez in "Uses and Abuses of the Stereotype Mechanism in UML 1.x and 2.0". == Diagrams == UML 2 defines many types of diagrams – shown as a taxonomy in the image. === Structure diagrams === Structure diagrams emphasize the structure of the system – using objects, classifiers, relationships, attributes and operations. They are used to document software architecture. Class diagram – Describes the structure of a class Component diagram – Describes how a software system is split into components and dependencies between the components Composite structure diagram Deployment diagram Object diagram Package diagram Profile diagram === Behavior diagrams === Behavior diagrams emphasize the behavior of a system by showing collaborations among objects and changes to the internal states of objects. They are used to describe the functionality of a system. Activity diagram – Describes the business and operational activities of components State machine diagram Use case diagram – Depicts of a user's interaction with a system === Interaction diagrams === Interaction diagrams, a subset of behavior diagrams, emphasize the flow of control and data between components of a system. Communication diagram – shows communication between components Interaction overview diagram Sequence diagram – shows interactions arranged in time sequence; can be drawn via tools such as Lucidchart and Draw.io Timing diagram – focuses on timing constraints === Examples === == Adoption == In 2013, UML had been marketed by OMG for many contexts, but aimed primarily at software development with limited success. It has been treated, at times, as a design silver bullet, which leads to problems. UML misuse includes overuse (designing every part of the system with it, which is unnecessary) and assuming that novices can design with it. It is considered a large language, with many constructs. Some people (including Jacobson) feel that UML's size hinders learning and therefore uptake. Visual Studio removed support for UML in 2016 due to lack of use. == History == UML has evolved since the second half of the 1990s and has its roots in the object-oriented programming methods developed in the late 1980s and early 1990s. The image shows a timeline of the history of UML and other object-oriented modeling methods and notation. === Origin === Rational Software hired James Rumbaugh from General Electric in 1994 and after that, the company became the source for two of the most popular object-oriented modeling approaches of the day: Rumbaugh's object-modeling technique (OMT) and Grady Booch's method. They were soon assisted in their efforts by Ivar Jacobson, the creator of the object-oriented software engineeri
Read more →
Environmental impact of AI

The environmental impact of the design, training, deployment and use of artificial intelligence includes the greenhouse gas emissions from generating electricity for data centres and computing hardware, operational and upstream water use, and material impacts from hardware manufacturing, mining and electronic waste. Estimating AI's environmental effects can be difficult because results depend on how impacts are measured, including whether accounting includes only model computation or also data-centre overhead, idle capacity, hardware manufacture, and local electricity supply. As these issues have received greater attention, governments and regulators have increasingly considered data-centre reporting requirements, energy-efficiency standards, and broader transparency measures for AI-related resource use. == Carbon footprint and energy use == AI-related energy use arises at multiple stages, including model training, fine-tuning, inference, storage, networking, and supporting infrastructure such as cooling and power conversion. === Individual level === Published estimates of energy use per AI request vary widely across models, tasks and measurement methods. A benchmark study presented at the 2024 ACM Conference on Fairness, Accountability, and Transparency found substantial differences between task types, with lower energy use for some text tasks and much higher energy use for image generation in the study's test conditions. In that benchmark, simple classification tasks consumed about 0.002–0.007 Wh per prompt on average (about 9% of a smartphone charge for 1,000 prompts), while text generation and text summarisation each used about 0.05 Wh per prompt; image generation averaged 2.91 Wh per prompt, and the least efficient image model in the study used 11.49 Wh per image (roughly equivalent to half a smartphone charge). First-party measurements in production environments have also been published. A 2025 Google study on Gemini assistant serving reported median per-prompt energy, emissions, and water-use estimates under the authors' accounting framework, while noting that different system boundaries can produce substantially different results. The study reported a median text-prompt estimate of about 0.24 Wh, which is roughly as much energy as watching nine seconds of television. The study also stated that software and infrastructure improvements reduced energy use by a factor of 33 and carbon emissions by a factor of 44 for a typical prompt over one year within the authors' framework. Researchers at the University of Michigan measured the energy consumption of various Meta Llama 3.1 models released in 2024 and found that smaller language models (8 billion parameters) use about 114 joules (0.03167 Wh) per response, while larger models (405 billion parameters) require up to 6,700 joules (1.861 Wh) per response. This corresponds to the energy needed to run a microwave oven for roughly one-tenth of a second and eight seconds, respectively. Comparisons between AI systems and human labour for specific tasks have produced mixed results and remain sensitive to assumptions about output quality, workload and system boundaries. A 2024 study in Scientific Reports reported 130 to 2900 times lower estimated carbon emissions for selected AI systems than for human writers and illustrators under its assumptions. A later Scientific Reports paper reported a counterexample for programming tasks under its assumptions, finding 5 to 19 times higher estimated emissions for the evaluated AI system than for human programmers on the benchmark used in that study. === System level === ==== Energy use and efficiency ==== AI electricity intensity depends not only on model architecture but also on hardware and facility efficiency. Data-centre operators commonly report Power usage effectiveness (PUE), which measures the ratio of total facility energy to IT equipment energy; a lower PUE indicates less overhead energy for cooling and other supporting infrastructure. Operators may also publish metrics and case studies on hardware efficiency, cooling systems and power sourcing. In its 2024 environmental report, Google stated that its 2023 total greenhouse gas emissions increased 13% year over year, primarily because of increased data-centre energy consumption and supply-chain emissions, while also reporting lower PUE than industry averages for its own facilities. The International Energy Agency has also reported that data centres remain a relatively small share of global electricity use overall, but that their local effects can be much more pronounced because demand is geographically concentrated. ==== Carbon footprint ==== At system level, AI contributes to rising electricity demand in data centres and related infrastructure. The International Energy Agency estimated that data centres used about 415 TWh of electricity in 2024, or around 1.5% of global electricity consumption, and projected that data-centre electricity use could rise to about 945 TWh by 2030, with AI identified as the main driver of that growth alongside other digital services. The carbon footprint of AI systems depends strongly on electricity sources, hardware efficiency, utilisation rates, and what stages are included in the accounting. Training large models can require substantial electricity, while total lifecycle impacts also depend on deployment scale and the amount of inference performed after training. Early analyses of frontier-model development reported rapid historical growth in training compute for selected systems, although later trends have depended on changes in model design, hardware and efficiency gains. Accounting methods that include upstream or embodied impacts, such as hardware manufacture and facilities construction, can materially affect estimates of AI-related emissions. === Decisions and strategies by individual companies === Large technology companies have reported that the expansion of AI and cloud infrastructure affects their sustainability targets, electricity demand, and resource use. Google, for example, attributed part of its emissions growth in 2023 to increased data-centre energy consumption and supply-chain emissions in its 2024 environmental report. Cloud and AI companies have also announced measures intended to reduce environmental impacts, including investment in more efficient hardware, low-carbon electricity procurement, alternative cooling systems, and water stewardship programmes. The extent, comparability, and third-party verification of such disclosures vary between firms and jurisdictions. == Water usage == Data centres can use water directly for cooling and indirectly through the water used in electricity generation, depending on the local energy mix. Public reporting on data-centre water use has often been inconsistent, making comparisons between operators and regions difficult. To standardise operational reporting, The Green Grid proposed the metric water usage effectiveness (WUE), defined as annual site water use divided by IT equipment energy use. WUE does not by itself measure local water stress, source sustainability, or all upstream water impacts. Studies of AI water use also distinguish between water withdrawal and water consumption. Research on AI-specific water use has argued that the water footprint of AI systems can be difficult to observe and may vary substantially by location, cooling design, and electricity source. A 2025 Communications of the ACM article summarised methods for estimating AI water footprints and emphasised the distinction between water withdrawal and water consumption. Li and colleagues estimated that global AI water withdrawal could reach 4.2–6.6 billion cubic metres in 2027 under the scenarios examined in their article. Using GPT-3, released by OpenAI in 2020, as an example, they estimated that training the model in Microsoft's U.S. data centres could consume about 700,000 litres of onsite water and about 5.4 million litres in total when offsite electricity-related water use was included; they also estimated that 10–50 medium-length GPT-3 responses could consume about 500 mL of water, depending on when and where the model was deployed. Published prompt-level estimates have also varied by system and accounting framework: the 2025 Google study on Gemini assistant serving reported a median text-prompt estimate of about 0.26 mL under its framework. Location can materially affect the significance of data-centre water use. Research on U.S. data centres found that one-fifth of servers' direct water footprint came from moderately to highly water-stressed watersheds, while nearly half of servers were fully or partially powered by plants located in water-stressed regions. A 2025 Reuters report, citing data from Verisk Maplecroft and NatureFinance, said that an average mid-sized data centre uses about 1.4 million litres of water per day for cooling and that Phoenix would experience a 32% increase in annual water stress if currently pl
Read more →
TasteDive

TasteDive (formerly named TasteKid) is an entertainment recommendation engine for films, TV shows, music, video games, books, people, places, and brands. It also has elements of a social media site; it allows users to connect with "tastebuds", people with like minded interests. == History == TasteDive was founded in 2008 as TasteKid by brothers Andrei Oghina and Felix Oghina. In 2019, it was acquired by Qloo headquartered in NYC. "Qloo has built for developers and enterprises what TasteDive has built for individuals". == Description == When a user types in the title of a film or TV show, the site's algorithm provides a list of similar content. It provides recommendations for TV shows to watch based on films liked by the user, and vice versa. It also provides recommendations for music, video games, and books, and includes film and TV trailers and music videos. An account is free and is not required to receive recommendations, but recommendations are more accurate for those with an account. The more a user explores the site, the more the site learns about the user's preferences and the better the results become. The site also has a social media aspect where one can see activity and gain recommendations from other users, how many others in the community like or dislike any recommendation, and how popular their tastes are within the TasteDive community. The main competitors of TasteDive are Taste App, Trakt.tv and Tastoid.
Read more →
RuleML

RuleML is a global initiative, led by a non-profit organization RuleML Inc., that is devoted to advancing research and industry standards design activities in the technical area of rules that are semantic and highly inter-operable. The standards design takes the form primarily of a markup language, also known as RuleML. The research activities include an annual research conference, the RuleML Symposium, also known as RuleML for short. Founded in fall 2000 by Harold Boley, Benjamin Grosof, and Said Tabet, RuleML was originally devoted purely to standards design, but then quickly branched out into the related activities of coordinating research and organizing an annual research conference starting in 2002. The M in RuleML is sometimes interpreted as standing for Markup and Modeling. The markup language was developed to express both forward (bottom-up) and backward (top-down) rules in XML for deduction, rewriting, and further inferential-transformational tasks. It is defined by the Rule Markup Initiative, an open network of individuals and groups from both industry and academia that was formed to develop a canonical Web language for rules using XML markup and transformations from and to other rule standards/systems. Markup standards and initiatives related to RuleML include: Rule Interchange Format (RIF): The design and overall purpose of W3C's Rule Interchange Format (RIF) industry standard is based primarily on the RuleML industry standards design. Like RuleML, RIF embraces a multiplicity of potentially useful rule dialects that nevertheless share common characteristics. RuleML Technical Committee from Oasis-Open: An industry standards effort devoted to legal automation utilizing RuleML. Semantic Web Rule Language (SWRL): An industry standards design, based primarily on an early version of RuleML, whose development was funded in part by the DARPA Agent Markup Language (DAML) research program. Semantic Web Services Framework, particularly its Semantic Web Services Language: An industry standards design, based primarily on a medium-mature version of RuleML, whose development was funded in part by the DARPA Agent Markup Language (DAML) research program and the WSMO research effort of the EU. Mathematical Markup Language (MathML): However, MathML's Content Markup is better suited for defining functions rather than relations or general rules Predictive Model Markup Language (PMML): With this XML-based language one can define and share various models for data-mining results, including association rules Attribute Grammars in XML (AG-markup): For AG's semantic rules, there are various possible XML markups that are similar to Horn-rule markup Extensible Stylesheet Language Transformations (XSLT): This is a restricted term-rewriting system of rules, written in XML, for transforming XML documents into other text documents
Read more →
Learning rule

An artificial neural network's learning rule or learning process is a method, mathematical logic or algorithm which improves the network's performance and/or training time. Usually, this rule is applied repeatedly over the network. It is done by updating the weight and bias levels of a network when it is simulated in a specific data environment. A learning rule may accept existing conditions (weights and biases) of the network, and will compare the expected result and actual result of the network to give new and improved values for the weights and biases. Depending on the complexity of the model being simulated, the learning rule of the network can be as simple as an XOR gate or mean squared error, or as complex as the result of a system of differential equations. The learning rule is one of the factors which decides how fast or how accurately the neural network can be developed. Depending on the process to develop the network, there are three main paradigms of machine learning: supervised learning, unsupervised learning, and reinforcement learning. == Background == A lot of the learning methods in machine learning work similar to each other, and are based on each other, which makes it difficult to classify them in clear categories. But they can be broadly understood in 4 categories of learning methods, though these categories don't have clear boundaries and they tend to belong to multiple categories of learning methods - Hebbian - Neocognitron, Brain-state-in-a-box Gradient Descent - ADALINE, Hopfield Network, Recurrent Neural Network Competitive - Learning Vector Quantisation, Self-Organising Feature Map, Adaptive Resonance Theory Stochastic - Boltzmann Machine, Cauchy Machine Though these learning rules might appear to be based on similar ideas, they do have subtle differences, as they are a generalisation or application over the previous rule, and hence it makes sense to study them separately based on their origins and intents. === Hebbian Learning === Developed by Donald Hebb in 1949 to describe biological neuron firing. In the mid-1950s it was also applied to computer simulations of neural networks. Δ w i = η x i y {\displaystyle \Delta w_{i}=\eta x_{i}y} Where η {\displaystyle \eta } represents the learning rate, x i {\displaystyle x_{i}} represents the input of neuron i, and y is the output of the neuron. It has been shown that Hebb's rule in its basic form is unstable. Oja's Rule, BCM Theory are other learning rules built on top of or alongside Hebb's Rule in the study of biological neurons. ==== Perceptron Learning Rule (PLR) ==== The perceptron learning rule originates from the Hebbian assumption, and was used by Frank Rosenblatt in his perceptron in 1958. The net is passed to the activation (transfer) function and the function's output is used for adjusting the weights. The learning signal is the difference between the desired response and the actual response of a neuron. The step function is often used as an activation function, and the outputs are generally restricted to -1, 0, or 1. The weights are updated with w new = w old + η ( t − o ) x i {\displaystyle w_{\text{new}}=w_{\text{old}}+\eta (t-o)x_{i}} where "t" is the target value and "o" is the output of the perceptron, and η {\displaystyle \eta } is called the learning rate. The algorithm converges to the correct classification if: the training data is linearly separable η {\displaystyle \eta } is sufficiently small (though smaller η {\displaystyle \eta } generally means a longer learning time and more epochs) It should also be noted that a single layer perceptron with this learning rule is incapable of working on linearly non-separable inputs, and hence the XOR problem cannot be solved using this rule alone === Backpropagation === Seppo Linnainmaa in 1970 is said to have developed the Backpropagation Algorithm but the origins of the algorithm go back to the 1960s with many contributors. It is a generalisation of the least mean squares algorithm in the linear perceptron and the Delta Learning Rule. It implements gradient descent search through the space possible network weights, iteratively reducing the error, between the target values and the network outputs. ==== Widrow-Hoff Learning (Delta Learning Rule) ==== Similar to the perceptron learning rule but with different origin. It was developed for use in the ADALINE network, which differs from the Perceptron mainly in terms of the training. The weights are adjusted according to the weighted sum of the inputs (the net), whereas in perceptron the sign of the weighted sum was useful for determining the output as the threshold was set to 0, -1, or +1. This makes ADALINE different from the normal perceptron. Delta rule (DR) is similar to the Perceptron Learning Rule (PLR), with some differences: Error (δ) in DR is not restricted to having values of 0, 1, or -1 (as in PLR), but may have any value DR can be derived for any differentiable output/activation function f, whereas in PLR only works for threshold output function Sometimes only when the Widrow-Hoff is applied to binary targets specifically, it is referred to as Delta Rule, but the terms seem to be used often interchangeably. The delta rule is considered to a special case of the back-propagation algorithm. Delta rule also closely resembles the Rescorla-Wagner model under which Pavlovian conditioning occurs. === Competitive Learning === Competitive learning is considered a variant of Hebbian learning, but it is special enough to be discussed separately. Competitive learning works by increasing the specialization of each node in the network. It is well suited to finding clusters within data. Models and algorithms based on the principle of competitive learning include vector quantization and self-organizing maps (Kohonen maps).
Read more →
Microsoft Whiteboard

Microsoft Whiteboard is a free multi-platform application, as well as an online service and a feature in Microsoft Teams, which simulates a virtual whiteboard and enables real-time collaboration between users. == Overview and features == Microsoft Whiteboard allows users to draw on a virtual whiteboard using input methods such as a stylus pen or a mouse and keyboard, and write down notes, draw connections between shareable ideas, and interact in real time. Microsoft Whiteboard is available to download on the following platforms and devices: Microsoft Windows (on Windows 10 or above) Android Apple iOS Surface Hub devices It is also available on the web and as a feature in Microsoft Teams. Microsoft Whiteboard allows users with Microsoft accounts to view, edit, and share whiteboards using the provided tools and options. The feature set includes tools for drawing, shapes, and media. Drawing in Microsoft Whiteboard is called inking. It works both on mobile devices and computers. The inking toolbar has customizable pencils, a ruler, a highlighter, an eraser, and an object selector. Whiteboard can recognize shapes drawn by hand and straighten them. Holding the Shift key on a computer while inking draws straight lines. Microsoft Whiteboard has keyboard shortcuts for some functions. Additional features include inserting sticky notes, text boxes, stickers, as well as images. Grid lines and colors are adjustable. Different templates can be inserted into the whiteboard. Users can also share their reactions. A feature limited to boards created in Microsoft Teams, is the ability to make them read-only; other participants from the meeting cannot edit them. == Reviews == PC Magazine gave Microsoft Whiteboard a score of 3.5 out of 5, praising the app's free availability and plentiful templates. It compared it to other, paid whiteboarding solutions, and concluded that Microsoft offers the best free one. Some of the cons, described by PCMag, include the inability to view boards without a Microsoft account and the inability to create custom templates.
Read more →
LightGBM

LightGBM, short for Light Gradient-Boosting Machine, is a free and open-source distributed gradient-boosting framework for machine learning, originally developed by Microsoft. It is based on decision tree algorithms and used for ranking, classification and other machine learning tasks. The development focus is on performance and scalability. == Overview == The LightGBM framework supports different algorithms including GBT, GBDT, GBRT, GBM, MART and RF. LightGBM has many of XGBoost's advantages, including sparse optimization, parallel training, multiple loss functions, regularization, bagging, and early stopping. A major difference between the two lies in the construction of trees. LightGBM does not grow a tree level-wise — row by row — as most other implementations do. Instead it grows trees leaf-wise. It will choose the leaf with max delta loss to grow. Besides, LightGBM does not use the widely used sorted-based decision tree learning algorithm, which searches the best split point on sorted feature values, as XGBoost or other implementations do. Instead, LightGBM implements a highly optimized histogram-based decision tree learning algorithm, which yields great advantages on both efficiency and memory consumption. The LightGBM algorithm utilizes two novel techniques called Gradient-Based One-Side Sampling (GOSS) and Exclusive Feature Bundling (EFB) which allow the algorithm to run faster while maintaining a high level of accuracy. LightGBM works on Linux, Windows, and macOS and supports C++, Python, R, and C#. The source code is licensed under MIT License and available on GitHub. == Gradient-based one-side sampling == When using gradient descent, one thinks about the space of possible configurations of the model as a valley, in which the lowest part of the valley is the model which most closely fits the data. In this metaphor, one walks in different directions to learn how much lower the valley becomes. Typically, in gradient descent, one uses the whole set of data to calculate the valley's slopes. However, this commonly used method assumes that every data point is equally informative. By contrast, Gradient-Based One-Side Sampling (GOSS), a method first developed for gradient-boosted decision trees, does not rely on the assumption that all data are equally informative. Instead, it treats data points with smaller gradients (shallower slopes) as less informative by randomly dropping them. This is intended to filter out data which may have been influenced by noise, allowing the model to more accurately model the underlying relationships in the data. == Exclusive feature bundling == Exclusive feature bundling (EFB) is a near-lossless method to reduce the number of effective features. In a sparse feature space many features are nearly exclusive, implying they rarely take nonzero values simultaneously. One-hot encoded features are a perfect example of exclusive features. EFB bundles these features, reducing dimensionality to improve efficiency while maintaining a high level of accuracy. The bundle of exclusive features into a single feature is called an exclusive feature bundle.
Read more →
OpenAI Five

OpenAI Five is a computer program by OpenAI that plays the five-on-five video game Dota 2. Its first public appearance occurred in 2017, where it was demonstrated in a live one-on-one game against the professional player Dendi, who lost to it. The following year, the system had advanced to the point of performing as a full team of five, and began playing against and showing the capability to defeat professional teams. By choosing a game as complex as Dota 2 to study machine learning, OpenAI thought they could more accurately capture the unpredictability and continuity seen in the real world, thus constructing more general problem-solving systems. The algorithms and code used by OpenAI Five were eventually borrowed by another neural network in development by the company, one which controlled a physical robotic hand. OpenAI Five has been compared to other similar cases of artificial intelligence (AI) playing against and defeating humans, such as AlphaStar in the video game StarCraft II, AlphaGo in the board game Go, Deep Blue in chess, and Watson on the television game show Jeopardy!. == History == Development on the algorithms used for the bots began in November 2016. OpenAI decided to use Dota 2, a competitive five-on-five video game, as a base due to it being popular on the live streaming platform Twitch, having native support for Linux, and had an application programming interface (API) available. Before becoming a team of five, the first public demonstration occurred at The International 2017 in August, the annual premiere championship tournament for the game, where Dendi, a Ukrainian professional player, lost against an OpenAI bot in a live one-on-one matchup. After the match, CTO Greg Brockman explained that the bot had learned by playing against itself for two weeks of real time, and that the learning software was a step in the direction of creating software that can handle complex tasks "like being a surgeon". OpenAI used a methodology called reinforcement learning, as the bots learn over time by playing against itself hundreds of times a day for months, in which they are rewarded for actions such as killing an enemy and destroying towers. By June 2018, the ability of the bots expanded to play together as a full team of five and were able to defeat teams of amateur and semi-professional players. At The International 2018, OpenAI Five played in two games against professional teams, one against the Brazilian-based paiN Gaming and the other against an all-star team of former Chinese players. Although the bots lost both matches, OpenAI still considered it a successful venture, stating that playing against some of the best players in Dota 2 allowed them to analyze and adjust their algorithms for future games. The bots' final public demonstration occurred in April 2019, where they won a best-of-three series against The International 2018 champions OG at a live event in San Francisco. A four-day online event to play against the bots, open to the public, occurred the same month. There, the bots played in 42,729 public games, winning 99.4% of those games. == Architecture == Each OpenAI Five bot is a neural network containing a single layer with a 4096-unit LSTM that observes the current game state extracted from the Dota developer's API. The neural network conducts actions via numerous possible action heads (no human data involved), and every head has meaning. For instance, the number of ticks to delay an action, what action to select – the X or Y coordinate of this action in a grid around the unit. In addition, action heads are computed independently. The AI system observes the world as a list of 20,000 numbers and takes an action by conducting a list of eight enumeration values. Also, it selects different actions and targets to understand how to encode every action and observe the world. OpenAI Five has been developed as a general-purpose reinforcement learning training system on the "Rapid" infrastructure. Rapid consists of two layers: it spins up thousands of machines and helps them 'talk' to each other and a second layer runs software. By 2018, OpenAI Five had played around 180 years worth of games in reinforcement learning running on 256 GPUs and 128,000 CPU cores, using Proximal Policy Optimization, a policy gradient method. == Comparisons with other game AI systems == Prior to OpenAI Five, other AI versus human experiments and systems have been successfully used before, such as Jeopardy! with Watson, chess with Deep Blue, and Go with AlphaGo. In comparison with other games that have used AI systems to play against human players, Dota 2 differs as explained below: Long run view: The bots run at 30 frames per second for an average match time of 45 minutes, which results in 80,000 ticks per game. OpenAI Five observes every fourth frame, generating 20,000 moves. By comparison, chess usually ends before 40 moves, while Go ends before 150 moves. Partially observed state of the game: Players and their allies can only see the map directly around them. The rest of it is covered in a fog of war which hides enemies units and their movements. Thus, playing Dota 2 requires making inferences based on this incomplete data, as well as predicting what their opponent could be doing at the same time. By comparison, Chess and Go are "full-information games", as they do not hide elements from the opposing player. Continuous action space: Each playable character in a Dota 2 game, known as a hero, can take dozens of actions that target either another unit or a position. The OpenAI Five developers allow the space into 170,000 possible actions per hero. Without counting the perpetual aspects of the game, there are an average of ~1,000 valid actions each tick. By comparison, the average number of actions in chess is 35 and 250 in Go. Continuous observation space: Dota 2 is played on a large map with ten heroes, five on each team, along with dozens of buildings and non-player character (NPC) units. The OpenAI system observes the state of a game through developers' bot API, as 20,000 numbers that constitute all information a human is allowed to get access to. A chess board is represented as about 70 lists, whereas a Go board has about 400 enumerations. == Reception == OpenAI Five have received acknowledgement from the AI, tech, and video game community at large. Microsoft founder Bill Gates called it a "big deal", as their victories "required teamwork and collaboration". Chess champion Garry Kasparov, who lost against the Deep Blue AI in 1997, stated that despite their losing performance at The International 2018, the bots would eventually "get there, and sooner than expected". In a conversation with MIT Technology Review, AI experts also considered OpenAI Five system as a significant achievement, as they noted that Dota 2 was an "extremely complicated game", so even beating non-professional players was impressive. PC Gamer wrote that their wins against professional players was a significant event in machine learning. In contrast, Motherboard wrote that the victory was "basically cheating" due to the simplified hero pools on both sides, as well as the fact that bots were given direct access to the API, as opposed to using computer vision to interpret pixels on the screen. The Verge wrote that the bots were evidence that the company's approach to reinforcement learning and its general philosophy about AI was "yielding milestones". In 2019, DeepMind unveiled a similar bot for StarCraft II, AlphaStar. Like OpenAI Five, AlphaStar used reinforcement learning and self-play. The Verge reported that "the goal with this type of AI research is not just to crush humans in various games just to prove it can be done. Instead, it's to prove that — with enough time, effort, and resources — sophisticated AI software can best humans at virtually any competitive cognitive challenge, be it a board game or a modern video game." They added that the DeepMind and OpenAI victories were also a testament to the power of certain uses of reinforcement learning. It was OpenAI's hope that the technology could have applications outside of the digital realm. In 2018, they were able to reuse the same reinforcement learning algorithms and training code from OpenAI Five for Dactyl, a human-like robot hand with a neural network built to manipulate physical objects. In 2019, Dactyl solved the Rubik's Cube.
Read more →