Top-down design - Top-down design is a design approach that begins with broad goals and major concepts, then progressively refines them into more detailed structures and...
Top-k accuracy - Top-k accuracy is the proportion of evaluated examples for which the target label appears among the k highest-ranked predictions.
Top-k sampling - Top-k sampling is a text-generation method that limits each next-token draw to the k highest-scoring candidate tokens.
Top-p sampling - Top-p sampling, also called nucleus sampling, generates text by sampling from the smallest leading set of next-token candidates whose cumulative probability...
Topic-specific PageRank - Topic-specific PageRank is a PageRank distribution biased toward a chosen topic by changing where the modeled random surfer teleports.
Topological data analysis - Topological data analysis, or TDA, uses mathematical ideas about shape and connectivity to study data.
Topological deep learning - Topological deep learning develops neural models that use structured relationships described by topological or higher-order mathematical objects.
Topological sort - A topological sort is an ordering of every vertex in a directed acyclic graph such that each directed edge goes from an earlier vertex to a later one.
Torus interconnect - A torus interconnect is a network topology in which nodes are arranged along multiple coordinate dimensions with connections that wrap around each dimension.
Total correlation - Total correlation measures statistical dependence among several random variables by comparing the sum of their individual entropies with their joint entropy.
Total functions - Total functions assign exactly one output to every input in their stated domain.
Total operating characteristic - The total operating characteristic, or TOC, is a curve for evaluating how a ranked score distinguishes a binary reference outcome across thresholds.
Total Turing Test - The total Turing Test is an expanded behavioral test that includes perceptual and physical interaction as well as conversation.
Total variation - Total variation measures accumulated change across a specified structure.
Total Variation Distance (Tvd) - Total variation distance, usually abbreviated TVD or TV, measures how different two probability distributions are.
Totally Random Trees Embedding - Totally random trees embedding is an unsupervised method that represents an input by the leaves it reaches in a collection of randomly constructed trees.
Tournament - A tournament is a directed graph with exactly one directed edge between each pair of distinct vertices.
Tower - A tower is a substantial neural-network component within a larger neural network.
Toxicity - In AI evaluation, toxicity refers to abusive, threatening, insulting or otherwise harmful conversational content as defined by a particular assessment scheme.
Toxicity detection - Toxicity detection is the use of computational methods to identify language that meets a defined criterion for harmful, abusive, disrespectful, or...
Trace - The trace of a square matrix is the sum of its main-diagonal entries, written tr(A). Those entries run from the upper-left corner to the lower-right corner.
Traceability - Traceability is the ability to follow relevant connections between an AI system's inputs, data, processes, versions, outputs, and decisions through recorded...
Tracking signal - A tracking signal is a forecast-monitoring statistic highlighting persistent errors that may indicate bias or a changed forecasting environment.
Tractability - Tractability describes whether a computational problem can be solved with manageable resources.
Tragedy of the commons - The tragedy of the commons is a collective-action problem in which individually attractive use of a shared, limited resource can produce excessive use and a...
Trail - A trail is a record of reversible changes made during a search or computation, used to restore an earlier state when backtracking.
Train/validation/test split - A train/validation/test split divides a dataset into parts with different roles: training fits a model, validation guides development choices, and testing...
Trained model - A trained model is the result of model training: a model whose adjustable state has been fitted using a training process.
Training - Training is the process of adjusting a machine-learning model using data or experience.
Training curve - A training curve plots a model's measured performance against progress through training, such as optimization steps or epochs.
Training error - Training error is a model's measured loss on the examples used to fit it.
Training loss - Training loss is the value of the learning objective measured on examples used to fit a model.
Training oracle - A training oracle is a procedure that uses a known reference answer to provide guidance for a model’s training decisions.
Training Pipeline - A Training Pipeline is an organized sequence of steps that turns selected data and a training configuration into a fitted model and evidence about its...
Training-serving skew - Training-serving skew is a difference between how a model behaves or performs in its training and evaluation setting and how the same model behaves or...
Training set - A training set is the collection of examples used to fit a model's learned parameters.
Trajectory - In reinforcement learning, a trajectory is an ordered sequence of interactions between an agent and its environment.
Trajectory optimization - Trajectory optimization chooses a time-varying sequence of states and controls that minimizes a specified cost while satisfying dynamics and other constraints.
Transaction data - Transaction data records business activities or events, such as an order, a booking or a movement of goods.
Transaction logic - Transaction Logic is a formalism for describing and reasoning about state-changing actions within logic programming and databases.
Transcript-based editing - Transcript-based editing is a way to edit recorded speech and its associated media by selecting and arranging words in a time-linked transcript.
Transduction - Transduction is learning or inference directed at a particular supplied collection of cases, rather than producing a general prediction rule for arbitrary...
Transduction grammars - Transduction grammars are formal systems that describe correspondences between structured expressions, commonly between sentences in two languages.
Transfer-based machine translation - Transfer-based machine translation is an approach that analyses text in a source language, maps the resulting representation into a target-language...
Transfer function - A transfer function describes how an input is transformed into an output by a specified linear system, including delayed effects.
Transfer learning - Transfer learning reuses knowledge learned in one task or setting to help solve another.
Transfer model - A transfer model, in traditional machine translation, maps a representation of a source-language expression into a representation suitable for producing the...
Transform coding - Transform coding compresses data by first expressing it in a different mathematical representation, then encoding the resulting coefficients.