Gini impurity - Gini impurity measures how mixed the class labels are within a group of observations, commonly a node in a classification tree.
Gini's mean difference - Gini’s mean difference is a measure of dispersion defined as the average absolute difference between pairs of observations.
Gittins index - The Gittins index is a priority value for an individual option, or arm, in a classical discounted multi-armed bandit problem.
Given-new - Given-new describes a distinction in how information is presented in discourse.
Glejser test - The Glejser test is a regression diagnostic that investigates heteroscedasticity by relating the absolute residuals from a fitted model to selected...
GLIE - GLIE stands for greedy in the limit with infinite exploration.
Glitch token - A Glitch token is a token associated with unusually unreliable or anomalous behavior in a language model, such as difficulty reproducing its text or...
Global calibration - Global calibration, in the computer-experiment sense used by the NIST engineering-statistics glossary, means developing a function that transforms a...
Global constraint - A global constraint is a constraint that expresses a structured relationship over a collection of variables, often with a specialized method for propagating...
Global costmap - A global costmap is a spatial cost representation used to support planning a route across an operating area.
Global file system - A global file system presents a shared file namespace that users or applications can access from multiple locations, often across a wide-area network.
Global maximum - A global maximum is the greatest value that an objective function actually attains anywhere in its specified feasible set.
Global minimum - A global minimum is the smallest value an objective function actually reaches over its entire feasible set. A point that reaches it is a global minimizer.
Global optimization - Global optimization seeks the best objective value over the entire allowed set of choices.
Global serializability - Global serializability is the property that transactions spanning several databases or resources have an overall effect equivalent to some serial execution...
Glottal - Glottal means relating to the glottis, the opening between the vocal folds in the larynx.
Glottal stop - A glottal stop is a speech sound produced by a closure at the glottis, where the vocal folds come together and interrupt airflow.
Glyph - A glyph is a visual form used to represent written text or symbols.
Goal - A goal is a desired condition that an AI agent is intended to achieve or establish.
Goal clauses - Goal clauses are clauses with no positive literals, used in Horn-clause logic and resolution-based reasoning to represent goals to be proved by refutation.
Goal-directed reasoning - Goal-directed reasoning organizes inference around a particular question or target conclusion.
Goal formulation - Goal formulation is the process of deciding what an agent should try to achieve and expressing that outcome clearly enough to guide reasoning.
Goal monitoring - Goal monitoring is the process of checking whether the goals an agent is currently pursuing remain the most appropriate goals to pursue.
Goal test - A goal test is a procedure that determines whether a candidate state satisfies the goal of a search or planning problem.
Gödel machine - A Gödel machine is a theoretical self-improving problem solver that changes its own software only after finding a proof that the change is useful under its...
Gold labels - Gold labels are the reference annotations treated as the accepted answers for a defined machine-learning task or evaluation.
Gold standard - A gold standard is an accepted reference used to judge the correctness or quality of another result.
Golden record (informatics) - A golden record is a governed, consolidated representation of an entity that an organization treats as an authoritative reference for a defined purpose.
Golden response - A golden response is a carefully selected reference answer used to represent an expected or high-quality response to a particular input.
Golden-section search - Golden-section search is a derivative-free method for locating a minimum of a unimodal function on an interval.
Goldfeld - Quandt test - The Goldfeld - Quandt test is a test for heteroscedasticity that compares residual variance estimates from two separately fitted regression subsamples.
Golomb coding - Golomb coding is a family of lossless variable-length codes for nonnegative integers.
Gomory-Hu Tree - A Gomory-Hu tree is a weighted tree that compactly represents the minimum cut values between every pair of vertices in an undirected capacitated graph.
Gompertz distribution - The Gompertz distribution is a continuous distribution for nonnegative event times whose hazard changes exponentially with time.
Goodman and Kruskal's gamma - Goodman and Kruskal's gamma is a measure of ordinal association based on concordant and discordant pairs of observations.
Goodness-of-fit - Goodness of fit describes how well a statistical model agrees with the data being examined.
Goodnessoffit test - A goodness-of-fit test evaluates whether observed data are sufficiently inconsistent with a specified statistical model to provide evidence against it.
GOP (group of pictures) - A GOP, or group of pictures, is a grouping or prediction structure within compressed video, commonly organized around successive keyframe or refresh points.
Gorn address - A Gorn address identifies a node in an ordered rooted tree by recording the sequence of child choices needed to reach it from the root.
Gossip protocol - A gossip protocol spreads information through repeated exchanges among nodes in a network, often with peers selected partly at random.
Gradient - The gradient of a differentiable scalar-valued function is the vector of its partial derivatives.
Gradient accumulation - Gradient accumulation is a training technique that combines gradients from several smaller batches before performing one optimizer update.
Gradient bandit algorithm - A gradient bandit algorithm learns action preferences and converts them into action-selection probabilities, typically with a softmax function.
Gradient-based method - A gradient-based method is an optimization procedure that uses derivatives of an objective to guide its search.
Gradient boosted (decision) trees - Gradient boosted decision trees are an ensemble model built by adding decision trees in stages, with each new tree trained to improve the current ensemble...
Gradient boosting - Gradient boosting is a modeling method that builds a strong predictor from many small decision trees added one at a time, each new tree trained to correct...
Gradient Checkpointing - Gradient checkpointing, also commonly called activation checkpointing, is a training technique that reduces stored intermediate activations by recomputing...
Gradient clipping - Gradient clipping is a training technique that limits the magnitude of gradients before an optimizer uses them to update model parameters.