Thursday, July 23, 2026
- Image Classification Modelsv6Image classification models are machine learning systems that assign one or more category labels to a whole input image, the task that drove the modern wave of deep learning in computer vision.
- Individual Fairnessv5Individual fairness is the principle in machine learning that any two individuals who are similar with respect to a task should receive similar algorithmic outcomes.
- Image-to-Text Modelsv6Image-to-text models are machine learning systems that take an image as input and produce natural language text as output.
- Humanity's Last Examv10Humanity's Last Exam (HLE) is a multi-modal AI benchmark of 2,500 public expert-level academic questions (plus a 500-question private holdout, 3,000 in total) spanning more than 100 disciplines
- Imbalanced Datasetv6An imbalanced dataset is a dataset used in machine learning where the classification categories are not approximately equally represented, so that one class (the majority class) contains far more samples than…
- Hyperplanev6A hyperplane is a flat, affine subspace of dimension n-1 embedded in an n-dimensional space, defined by the linear equation $$w \cdot x + b = 0$$, where w is a normal vector and b is a scalar offset .
- Image-to-Image Modelsv5Image-to-image models (often shortened to img2img) are machine learning systems that take an input image and output a transformed version of it.
- IFBenchv4IFBench (Instruction Following Benchmark) is an artificial intelligence benchmark that measures whether large language models can follow precise output constraints they have never seen during training.
- Implicit Biasv7Implicit bias is an umbrella term that, in artificial intelligence and machine learning, refers to systematic tendencies operating below the surface of explicit design choices.
- IT Management ChatGPT Pluginsv4IT management ChatGPT plugins were a group of third party extensions for ChatGPT focused on tasks that information technology teams handle every day, including server administration, infrastructure…
- Incompatibility of Fairness Metricsv5The incompatibility of fairness metrics (also called the impossibility theorem of fairness or fairness trade-offs) is the proven mathematical result that several widely used definitions of algorithmic fairness…
- In-set conditionv4An in-set condition is a split condition used inside a decision tree node that tests whether the value of a single categorical feature belongs to a specified subset of that feature's possible values.
- Hierarchical Clusteringv7Hierarchical clustering is an unsupervised learning method that groups data into a tree of nested clusters, building the hierarchy by repeatedly merging the most similar groups (agglomerative, bottom-up) or…
- How to Steal ChatGPT-4, GPT-4 and other Proprietary LLMsv4This article is a defensive and academic survey of how proprietary large language models (LLMs) such as GPT-4, ChatGPT, Claude, and Gemini can be partially copied or have their internal information leaked…
- Gym (OpenAI Gym / Gymnasium)v8Gym, often written as OpenAI Gym, is an open source Python toolkit for developing and comparing reinforcement learning algorithms, originally released by openai on April 27, 2016.
- Ground Truthv7Ground truth is verified, correct information that serves as the authoritative reference for training and evaluating machine learning models.
- Health ChatGPT Pluginsv4Health ChatGPT plugins were a now-deprecated category of third-party tools that extended ChatGPT with capabilities for tracking nutrition, exercise, hydration, and habits, plus searching clinical trials and…
- HealthBench Hardv3HealthBench Hard is a 1,000 example curated subset of the HealthBench benchmark released by OpenAI on May 12, 2025.
- Guidesv7The Guides section of AI Wiki is a curated index of how-to articles, explainer pages, reading lists, and reference material that guide readers from a first encounter with artificial intelligence toward…
- Hashingv6Hashing in machine learning is the use of hash functions to map data of arbitrary size (a word, a URL, a document, a vector) into a fixed-size integer range, trading a small and controllable amount of…
- Heuristicv5A heuristic is a practical problem-solving approach that trades optimality, completeness, accuracy, or precision for speed, producing a good-enough answer when an exact method would be too slow, too expensive…
- Hidden Layerv10A hidden layer is a layer of artificial neurons in a neural network that sits between the input layer and the output layer
- Hinge Lossv7Hinge loss is the margin-based loss function defined as max(0, 1 - y * f(x)), used to train support vector machines (SVMs) and other maximum-margin classifiers, where y in {+1, -1} is the true label and f(x)…
- Healthv5AI in healthcare is the use of artificial intelligence, and especially machine learning, to diagnose disease, discover drugs, document care, predict patient risk, and run health operations.
- HellaSwagv9HellaSwag is a commonsense reasoning benchmark for language models, introduced by Rowan Zellers, Ari Holtzman, Yonatan Bisk, Ali Farhadi, and Yejin Choi in the 2019 paper "HellaSwag: Can a Machine Really…
- Group Attribution Biasv6Group attribution bias is the tendency to assume that what is true of one member of a group is true of the entire group
- How to Prevent OpenAI and Google From Training Their LLMs on Your Website's Datav4You can stop OpenAI, Google, Anthropic, and most other major AI companies from using your website to train large language models (LLMs) by adding a small set of directives to your site's robots.txt file.
- Holdout datav5Holdout data is a portion of a labeled dataset that is deliberately kept out of training so it can be used later to evaluate how well a model performs on examples it has never seen.
- Grok 3 Jailbreakv4Grok 3 jailbreak is the umbrella term for a class of reported guardrail-bypass findings against Grok 3, the third-generation large language model that xAI released on February 17, 2025.
- GeoAIv4GeoAI, short for geospatial artificial intelligence, is the application of artificial intelligence, especially machine learning and deep learning, to geographic data and spatial problems.
- Greedy Policyv6In reinforcement learning, a greedy policy is a decision rule that, in every state, selects the action with the highest estimated value, formally the action that maximizes the action-value function $$Q(s, a)$$…
- Gradient boosted (decision) trees (GBT)v7Gradient boosted decision trees (GBT, also written GBDT, GBM, or GBRT) is an ensemble method that builds a strong predictor by sequentially fitting many shallow decision trees to the negative gradient of a…
- Gradientv6In machine learning, the gradient is the vector of partial derivatives of a loss function with respect to every model parameter, and it points in the direction in which the loss increases most steeply.
- Generative Modelv8A generative model is a class of statistical and machine learning model that learns the joint probability distribution P(X) of the observed data, or the joint distribution P(X, Y) of inputs and labels
- Graphv4In machine learning the word graph has two unrelated meanings. The first is a graph as a data structure: a set of nodes (vertices) connected by edges, written formally as G = (V, E), used to represent…
- GeoBenchv4GeoBench is the umbrella name for a family of artificial intelligence benchmarks evaluating models on geospatial reasoning, earth monitoring, and geographic localization.
- Gini Impurityv6Gini impurity is the probability that a randomly chosen element from a dataset would be incorrectly classified if it were labeled at random according to the distribution of class labels in that dataset.
- Graph Executionv7Graph execution is a computation paradigm in machine learning frameworks where mathematical operations are organized into a directed acyclic graph (DAG) before being executed, so the runtime sees the whole…
- Graph Machine Learning Modelsv4Graph machine learning models are neural networks designed to operate on data structured as graphs, where the input is a set of nodes connected by edges rather than a grid like an image or a sequence like text.
- Geography ChatGPT Pluginsv4Geography ChatGPT Plugins were the subset of third party plugins published in the ChatGPT Plugins store between March 2023 and April 2024 that focused on maps, coordinates, satellite imagery, demographic and…
- Generatorv7A generator is a neural network within a generative adversarial network (GAN) that learns to produce synthetic data samples from random noise.
- Gradient clippingv6Gradient clipping is a training technique that caps the magnitude of gradient values before they update model weights
- Fully Connected Layerv6A fully connected layer (also called a dense layer or linear layer) is a layer in an artificial neural network in which every input value connects to every output value through a learned weight
- GPT APIv6The GPT API is the public HTTP interface that OpenAI exposes for programmatic access to its hosted language, vision, audio, image, and video models.
- GANv7GAN stands for generative adversarial network, a class of deep learning generative models in which two neural networks are trained against each other: a generator that fabricates synthetic data and a…
- Full Softmaxv7Full softmax (also called the standard softmax or exact softmax) is the softmax computation that calculates a probability for every possible output class in a classification model
- GPT-1v10GPT-1 is the first model in the GPT (Generative Pre-trained Transformer) series, a 117-million-parameter, 12-layer decoder-only Transformer released by OpenAI on June 11, 2018 in the paper "Improving Language…
- Feedback Loopv5A feedback loop in machine learning is a cycle in which a deployed model's predictions influence the real world, and the resulting data is then collected and used to retrain the same model
- Few-Shot Learningv9Few-shot learning is a branch of machine learning in which a model learns to recognize new classes or perform new tasks from only a very small number of labeled examples, typically one to five per class .
- Forget Gatev5The forget gate is a sigmoid layer inside a Long Short-Term Memory (LSTM) recurrent neural network that decides, element by element, how much of the previous cell state to keep and how much to discard at each…
- Fox (benchmark)v8Fox is an evaluation suite for fine-grained, multi-page document understanding by large vision-language models.
- Finance ChatGPT Pluginsv4Finance ChatGPT Plugins are a now-deprecated category of third-party extensions that, during 2023 and 2024, connected ChatGPT to live market data, regulatory feeds, payments rails, and personal finance tools.
- Fine-tune ChatGPT with Perplexity, Burstiness, Professionalism, Randomness and Sentimentality Guidev4The perplexity, burstiness, professionalism, randomness, and sentimentality guide is a prompt engineering pattern that asks ChatGPT to write according to five style sliders, each rated 1 to 10, so a user can…
- Food ChatGPT Pluginsv4Food ChatGPT Plugins were the subset of third-party plugins published in the ChatGPT plugins store between March 2023 and April 2024 that focused on cooking, recipes, grocery shopping, restaurant discovery…
- AI in the food industryv3AI in the food industry refers to the application of artificial intelligence, machine learning, computer vision, and robotics across the food and beverage value chain, from farming and processing through…
- Feature Setv6A feature set is the complete collection of input variables (features, attributes, or predictors) that a machine learning model uses to learn patterns and make predictions.
- Feature Extractionv6Feature extraction is the process of transforming raw data into a smaller set of derived, informative numerical variables called features that capture the essential structure of the input while discarding…
- Feature Importancesv6Feature importances are numeric scores that quantify how much each input feature contributes to the predictions of a machine learning model.
- Feature Vectorv9A feature vector is an n-dimensional, ordered list of numerical values that represents the measurable properties of an object, data point, or observation in a format suitable for processing by machine learning…
- Fill-Mask Modelsv5Fill-mask models are language models trained with a masked language modeling (MLM) objective, in which a fraction of the tokens in an input sequence are hidden behind a special [MASK] symbol and the model…
- Feedforward Neural Network (FFN)v12A feedforward neural network (FFN), also called a multilayer perceptron (MLP) when it has multiple layers, is a type of artificial neural network in which information flows in one direction only, from the…
- Feature Extraction Modelsv4Feature extraction models are machine learning systems that transform raw inputs such as text, images, or audio into dense numerical vectors known as embeddings or hidden-state representations.
- Estimatorv5An estimator is a rule, function, or algorithm that takes observed data and produces a value intended to approximate some unknown quantity, typically a parameter of a probability distribution or a function…
- Examplev5In machine learning, an example is a single data point that a model trains on or makes a prediction about: in supervised settings it is a pair $(x, y)$ where $x$ is a vector of features and $y$ is a label, and…
- Experimenter's Biasv5Experimenter's bias (also called the observer-expectancy effect, experimenter expectancy effect, or experimenter effect) is a type of cognitive bias in which a researcher's expectations or beliefs about the…
- False Negative Ratev6The false negative rate (FNR), also known as the miss rate, is the proportion of actual positive instances that a model or test incorrectly classifies as negative, computed as FNR = FN / (FN + TP).
- False positivev8A false positive (FP), also called a Type I error or a false alarm, is an instance whose true label is negative but whose predicted label is positive: the classifier raises an alarm where none was warranted.
- Factorio Learning Environmentv4The Factorio Learning Environment (FLE) is an open source benchmark and research framework that uses the industrial automation game Factorio to evaluate the long-horizon agentic capabilities of large language…
- Feature Engineeringv8Feature engineering is the process of using domain knowledge to create, transform, and select features from raw data so that machine learning models can learn more effectively.
- False negativev9A false negative (FN), also called a Type II error or a miss, is an instance whose true label is positive but that a classification model or test predicts as negative: a real positive case that the model fails…
- Fashion ChatGPT Pluginsv4Fashion ChatGPT Plugins were a small historical category of third-party plugins that connected ChatGPT to clothing, styling, and apparel services during the 2023 to 2024 plugin era run by OpenAI.
- Equalized Oddsv7Equalized odds is a group fairness criterion in machine learning that requires a classifier's true positive rate (TPR) and false positive rate (FPR) to be equal across all groups defined by a protected…
- Fairness Metricv7A fairness metric is a quantitative, mathematical measure used to evaluate whether a machine learning model's predictions or decisions treat different demographic groups equitably.
- False Positive Rate (FPR)v5The false positive rate (FPR) is the proportion of actual negative cases that a test, model, or decision process incorrectly classifies as positive, defined as FPR = FP / (FP + TN) where FP is the number of…
- Exploding Gradient Problemv12The exploding gradient problem is a training failure in deep neural networks where the gradients of the loss function grow exponentially large as they propagate backward through layers during backpropagation…
- Feature Crossv6A feature cross (also called a crossed feature or feature interaction) is a synthetic feature created by combining two or more existing features so that a model can learn their joint effect on a prediction.
- Fairness Constraintv6A fairness constraint is an explicit mathematical condition imposed on a machine learning model during training, evaluation, or post-processing that forces its predictions to satisfy a specified group-fairness…
- Experience Replayv7Experience replay is a reinforcement learning technique in which an agent stores its past transitions in a memory called a replay buffer and samples random mini-batches of those stored transitions to train on
- Featurev8In machine learning and statistics, a feature is an individual measurable property or characteristic of a phenomenon being observed, used as an input variable from which a model learns patterns and makes…
- Epochv12An epoch in machine learning is one complete pass through the entire training dataset, during which every example is presented to the model exactly once to compute gradients and update parameters before the…
- Educationv5In the AI sense, education refers to the use of artificial intelligence (especially large language models like ChatGPT) to teach, tutor, assess, and assist learning.
- Education ChatGPT Pluginsv7Education ChatGPT plugins were a category of third-party tools that extended ChatGPT with capabilities aimed at students, teachers, parents, and lifelong learners, and they are now deprecated.
- Epsilon Greedy Policyv5The epsilon-greedy policy (also written as ε-greedy) is a simple action-selection rule for reinforcement learning and multi-armed bandit problems: with probability 1 - ε the agent picks the action with the…
- Environment ChatGPT Pluginsv4Environment ChatGPT Plugins were a small, informal grouping of third party extensions for ChatGPT addressing sustainability, air quality, climate awareness, ecological purchasing, and related green living…
- Episode (Reinforcement Learning)v7An episode in reinforcement learning is one complete sequence of interaction between an agent and its environment, starting from an initial state and ending when a terminal state is reached.
- Entertainment ChatGPT Pluginsv4Entertainment ChatGPT Plugins were a now-deprecated category of third-party plugins that connected ChatGPT to streaming catalogs, recommendation engines, trivia banks, and audio playback services during the…
- Environmentv6In reinforcement learning (RL), an environment is the external system that an agent interacts with: it receives the agent's actions, transitions to a new state, and returns an observation together with a…
- Education Custom GPTsv5Education Custom GPTs are the Custom GPTs listed in the Education category of OpenAI's GPT Store: custom versions of ChatGPT built for teaching, tutoring, study help, and academic research.
- Entropyv6Entropy is a quantitative measure of the uncertainty, randomness, or average information content in a probability distribution: for a discrete random variable it equals the average number of yes/no questions…
- Encoderv8An encoder in machine learning is a neural network component that transforms input data (text, an image, audio, or code) into a compressed, structured representation, often called a latent representation…
- AI in entertainmentv4AI in entertainment refers to the use of artificial intelligence across the film, television, music, video game, animation, and live event industries.
- Early Stoppingv7Early stopping is a regularization technique that halts the training of an iterative machine learning model as soon as its performance on a held-out validation set stops improving
- Equality of Opportunityv5Equality of opportunity is a group-fairness criterion in machine learning that requires a classifier's true positive rate (TPR) to be equal across all groups defined by a sensitive attribute: qualified…
- Embedding Spacev6An embedding space is a continuous, typically high-dimensional vector space in which data objects (words, sentences, images, users, audio clips, code, or other entities) are represented as dense numerical…
- Embedding vectorv7An embedding vector is a dense, fixed-length array of real numbers that represents a discrete object (such as a word, sentence, image, audio clip, or graph node) as a point in a continuous vector space.
- Embedding Layerv9An embedding layer is a neural network component that acts as a trainable lookup table, mapping discrete integer indices (such as word IDs, user IDs, or category codes) to dense, continuous-valued vectors.
- Ensemblev5Ensemble methods are techniques in machine learning that combine the predictions of multiple models, known as base learners, to produce a single prediction that is typically more accurate and robust than any…
- Design ChatGPT Pluginsv4Design ChatGPT Plugins were third-party extensions inside ChatGPT that helped the chatbot create graphic assets, page layouts, logos, banners, and full websites without leaving the conversation, with named…
- Depthwise separable convolutional neural network (sepCNN)v4A depthwise separable convolutional neural network (often abbreviated sepCNN) is a convolutional neural network that replaces standard convolution layers with depthwise separable convolutions.
- Dynamicv4In machine learning, dynamic describes a model, a training process, or an inference process that runs frequently or continuously on fresh data, as opposed to one that runs once and stays fixed.
