B* - B* is a best-first tree-search procedure that seeks to establish which choice at the root is better than its alternatives.
B-frame - A B-frame is a video picture that supports bidirectional or bi-predictive coding, using reference information to predict image content efficiently.
B-rep - B-rep, short for boundary representation, describes a three-dimensional shape through its boundary elements and how they connect.
B-splines - B-splines are basis functions used to represent piecewise-polynomial curves and surfaces.
Back-propagation through time - Backpropagation through time, or BPTT, computes training gradients for a recurrent model by treating its successive time steps as an unfolded computational...
Back-testing - Back-testing evaluates a model, forecasting method or decision procedure using historical observations to examine how it would have performed.
Backbone - A backbone is the main feature-producing network within a larger model.
Backdoor - In machine learning, a backdoor is an intentionally introduced model behavior that is activated by a particular trigger or input condition.
Backed-up value - A backed-up value is a value assigned to an earlier state or search-tree node by combining values from its successors according to a specified decision rule.
Backfitting - Backfitting is an iterative method for fitting additive models: it cycles through the model's components, refitting each one to the residuals left by the...
Background knowledge - Background knowledge is information already available to an AI system that helps it interpret observations, reason about a domain or learn new relationships.
Background subtraction - Background subtraction is a computer-vision technique that identifies image regions differing from an estimated background scene.
Backjumping - Backjumping is a constraint-search technique that responds to a dead end by returning to an earlier decision relevant to the conflict, potentially skipping...
Backmarking - Backmarking is a constraint-search technique that remembers information from earlier compatibility checks so that the solver can avoid repeating checks...
Backoff - Backoff is a language-modeling strategy that uses a less specific estimate when the more specific context lacks sufficient evidence.
Backoff model - A backoff model uses a less specific model when the more specific context lacks sufficient evidence for a prediction.
Backpropagation - Backpropagation is an efficient method for calculating how a model's loss changes with respect to its parameters.
Backpropagation through structure - Backpropagation through structure is a method for calculating training gradients in recursive neural networks whose computations follow a structured input...
Backtracking line search - Backtracking line search is a procedure for choosing how far an optimization algorithm should move along a proposed direction.
Backtracking search - Backtracking search builds a candidate solution through a sequence of tentative choices.
Backtranslation - Backtranslation is a machine-translation training technique that turns target-language text into additional synthetic parallel data.
Backup - A backup is a retained copy of data, or a recoverable representation of it, created so information can be restored after loss, corruption, unwanted change...
Backup validation - Backup validation is the process of checking whether a backup meets defined requirements for integrity, completeness and intended recovery use.
Backus - Gilbert method - The Backus - Gilbert method is an approach to inverse problems that estimates local averages of an unknown quantity from indirect measurements.
Backus-Naur form - Backus-Naur form, or BNF, is a notation for specifying a context-free grammar.
Backward-chaining - Backward chaining is a goal-directed inference method that starts with a question and works backward through rules that could establish its answer.
Backward elimination - Backward elimination is a model-selection procedure that starts with a specified full set of candidate predictors and removes terms one at a time according...
Backward shift operator - The backward shift operator is a notation that moves a time-series value back by one observation period.
Bacterial colony optimization - Bacterial colony optimization is a population-based optimization method inspired by selected behaviors of bacteria.
BADIR - BADIR is Aryng's proprietary analytics framework for moving from a business question to a recommendation.
Bag of words - A bag of words is a way of representing text by counting which words appear in it and how often, ignoring grammar and word order.
Bagging - Bagging, short for bootstrap aggregating, is an ensemble method that trains many models on different random resamples of the training data and combines...
Balance - Balance, in the fairness sense used by the source glossary, requires a classifier to have approximately equal false-positive rates and false-negative rates...
Balanced accuracy score - Balanced accuracy score measures classification performance by averaging the recall of each class with equal class weight.
Ball divergence - Ball divergence is a statistical measure of how two probability distributions differ, constructed by comparing the probability they assign to collections of...
Ball tree - A ball tree is a data structure that organizes points into nested regions defined by a center and a radius.
Banding - Banding is the appearance of visible steps or stripes in a region that should change smoothly in brightness or color.
Bandit superprocess - A bandit superprocess is a sequential decision problem made from several independent Markov decision processes, with a shared restriction that only one...
Bar chart - A bar chart displays values using rectangular bars whose lengths represent magnitude on a shared numerical scale.
Barber-Agakov lower bound - The Barber-Agakov lower bound is a variational lower bound on mutual information that replaces an often difficult conditional distribution with a tractable...
Barlow Twins loss - Barlow Twins loss is a self-supervised learning objective that encourages two augmented views of the same data to produce consistent representations while...
Barnard exact test - The Barnard exact test is an unconditional exact procedure for comparing the success probabilities of two independent binomial samples, represented by a...
Barrier (computer science) - A barrier in concurrent computing is a synchronization point that coordinates a defined group of participants before they proceed.
Bartlett's test - Bartlett's test is a statistical test of whether several independent normal populations have the same variance.
Barycentric coordinates - Barycentric coordinates describe a point as a weighted combination of reference vertices.
Barzilai - Borwein method - The Barzilai - Borwein method is a gradient-based optimization approach that chooses its step size from the changes between two successive points and their...