Logistic regression - Logistic regression is a statistical and machine-learning method for modeling the probability of a categorical outcome.
Logit - The logit of a probability p is its natural log odds: logit(p) = ln[p/(1 − p)], for 0 < p < 1.
Logit lens - The logit lens is an interpretability technique that maps intermediate language-model representations into vocabulary scores using the model's output readout.
Logits - Logits are a classifier's raw output scores before they are converted into probabilities: the unbounded numbers the model actually computes, one per class...
Logrank test - The logrank test compares time-to-event distributions between groups while allowing right-censored observations.
Lomb-Scargle Periodograms - Lomb-Scargle periodograms assess periodic structure by fitting sinusoidal models across a range of frequencies, including when observations are unevenly...
Long-distance dependencies - Long-distance dependencies are relationships in which interpreting or predicting one part of a sequence requires information from a substantially separated...
Long-lived transaction - A long-lived transaction is a transaction whose duration is substantial relative to the system's usual transactions.
Long-range dependence - Long-range dependence is persistent statistical dependence across widely separated observations in a time series or another indexed process.
Long-running transaction - A long-running transaction is an extended unit of work whose completion may span multiple steps, services or waiting periods.
Long Short-Term Memory - Long Short-Term Memory, usually abbreviated LSTM, is a type of recurrent neural network designed to learn relationships across sequences.
Long tail - The long tail is the large collection of relatively infrequent items or events beyond a small group of highly popular ones in a ranked distribution.
Longitudinal data - Longitudinal data contain repeated observations of the same entities over time. The entities might be people, machines, stores, or accounts.
Longitudinal modified treatment policy - A longitudinal modified treatment policy, or LMTP, defines interventions over time that can modify the treatment value that would naturally arise at each...
Longitudinal or serial data - Longitudinal or serial data contain repeated measurements of a response on the same subject or unit over time.
Longitudinal study - A longitudinal study follows the same subjects or units over time, collecting observations that reveal change, continuity, and the ordering of events.
Longitudinal targeted maximum likelihood estimation - Longitudinal targeted maximum likelihood estimation, or LTMLE, is a framework for estimating specified causal quantities from data containing treatment...
Look-ahead (backtracking) - Look-ahead in backtracking is the use of current choices and constraints to examine remaining possibilities before extending a search branch further.
Look-up table - A look-up table, commonly called a LUT, stores output values that can be retrieved using an input value or combination of values.
Lookahead in regex - Lookahead in a regular expression checks whether a pattern does or does not match at the current position without consuming the characters inspected.
Loop dependence analysis - Loop dependence analysis examines relationships between operations in a loop to determine which execution-order constraints must be preserved.
Loop-level parallelism - Loop-level parallelism is the opportunity to execute work from multiple loop iterations concurrently while preserving the program's required behavior.
Loop scheduling - Loop scheduling determines how the iterations of a parallel loop are assigned to workers or threads.
Loop unrolling - Loop unrolling is a program transformation that expands a loop body to perform the work of several original iterations during one pass through the...
Loops - Loops are programming structures that repeat a block of instructions. A loop may visit each item in a collection or continue while a condition remains true.
Loopy belief propagation - Loopy belief propagation is the repeated application of belief-propagation message updates to a probabilistic graphical model containing cycles.
Loss curve - A loss curve is the plot of a model's error score over the course of training: the falling, or sometimes worryingly not-falling, line that shows whether and...
Loss function - A loss function assigns a numerical penalty to a model's output according to a specified objective.
Loss Of Situational Awareness Bias - Loss of situational awareness bias is a source-taxonomy term for a problem in which automation leaves people insufficiently aware of the current situation...
Loss surface - A loss surface is the landscape of a model's error over all possible parameter settings: the hills and valleys training must descend, whose shape decides...
Lossless compression - Lossless compression represents data in a form from which the defined original data can be recovered exactly.
Lossless join decomposition - Lossless join decomposition is the splitting of a relation into smaller relations so that joining their projections reconstructs exactly the original...
Lossless Join Property - The lossless join property means that a relation can be decomposed into smaller relations and reconstructed exactly by joining its projections.
Lossless predictive audio compression - Lossless predictive audio compression refers here to LPAC, a historical audio-coding system associated with Tilman Liebchen and research at the Technical...
Lossy compression - Lossy compression reduces the representation of data by allowing some information to be discarded or approximated.
Lossy data conversion - Lossy data conversion changes data from one representation to another in a way that prevents complete recovery of the original information from the...
Lost-in-the-middle effect - The lost-in-the-middle effect is the tendency of language models to use information at the beginning and end of a long input well, but to miss or garble...
Lot quality assurance sampling - Lot quality assurance sampling, or LQAS, is a sampling approach used to classify a lot, service area, or other defined group against quality or performance...
Loudness - Loudness is the perceived strength of a sound, from quiet to loud.
Low-dimensional embedding - A low-dimensional embedding represents an item using relatively few numerical coordinates while aiming to retain relationships useful for a task.
Low frame rate - Low frame rate in speech recognition means processing or emitting acoustic-model representations at fewer time steps per second than a reference configuration.
Low-pass filter - A low-pass filter preserves slowly varying components of a signal while reducing more rapidly varying components.
LU decomposition - LU decomposition factors a matrix into a lower-triangular matrix L and an upper-triangular matrix U, often after row permutations.
Lucas--Kanade method - The Lucas - Kanade method is a gradient-based approach to image alignment.
Lucky imaging - Lucky imaging captures many short exposures, selects those least affected by changing image degradation, aligns them, and combines them into an improved image.
Lusser's law - Lusser's law states that the reliability of a series system equals the product of its component reliabilities when component survival events are...
LZ4 (compression algorithm) - LZ4 is a lossless compression algorithm designed with an emphasis on fast compression and decompression.