underfitting
bias variance tradeoff, high bias model
Underfitting is what happens when a model is too simple, or hasn't trained long enough, to capture the real structure in its own training data — and the diagnostic is unambiguous…
224 technical terms and definitions
bias variance tradeoff, high bias model
Underfitting is what happens when a model is too simple, or hasn't trained long enough, to capture the real structure in its own training data — and the diagnostic is unambiguous…
balance, reduce
Undersampling is a technique for handling imbalanced datasets by reducing the number of majority class examples — rather than creating more minority examples (oversampling)…
signal & power integrity
Undershoot is a transient waveform excursion below the intended low logic level - It can forward-bias protection paths and introduce timing or reliability concerns. What Is…
underfitting, training convergence
Undertraining is the training condition where model has not received enough effective optimization or data exposure to realize its capacity - it leads to avoidable performance…
underwater robot autonomy, underwater robot, robot autonomy
Underwater Robot Autonomy Introduction & Motivation Underwater Robot Autonomy addresses a central problem in robotics process control and buoyancy-neutral-depth-control…
unicode standard, code point, UTF-8, UTF-16, UTF-32, Ken Thompson and Rob Pike, Unicode Consortium
Unicode is the universal standard that assigns a unique number, called a code point, to every character in the world's writing systems, together with the properties and rules…
nlp
Unicode normalization is the text canonicalization process that converts equivalent Unicode representations into a consistent standard form - it prevents hidden character-encoding…
Unidirectional attention restricts each token to attending only to previous tokens, enabling autoregressive generation. Mechanism: Causal mask applied, position i attends…
infrastructure
Unified memory is the shared virtual memory model that allows CPU and GPU to access a single logical address space - it simplifies programming by automating page migration, but…
managed memory allocation, page migration gpu, prefetching unified memory, memory oversubscription
Unified Memory is the CUDA programming model that provides a single memory address space accessible from both CPU and GPU — automatically migrating data between host and device…
unified virtual address space and translation overhead, unified virtual address, virtual address space, address space translation, space translation overhead, unified virtual
Unified Virtual Address Space and Translation Overhead Introduction & Motivation Unified Virtual Address Space and Translation Overhead addresses a central problem…
multimodal ai
Unified Vision-Language Models are architectures designed to process and generate both visual and textual data — tackling multiple tasks (VQA, captioning, retrieval, generation)…
continuous uniform, rectangular distribution, equal likelihood, random number generator, discrete uniform, inverse transform sampling
The uniform distribution assigns equal probability density to every value across a fixed range: for a continuous uniform variable on the interval from a to b, the density is…
uniformity, film thickness uniformity, deposition uniformity, cvd thickness uniformity, thickness non-uniformity, within wafer uniformity, wafer uniformity map, cvd uniformity calculation, cvd
Uniformity is a statistic, and the statistic that gets quoted decides which problems the fab is able to see. The number almost everyone reports is a half-range: the difference…
nlp
Unigram tokenization is the subword modeling approach that selects token segmentations by maximizing probability under a unigram language model over candidate pieces - it uses…
UJT, intrinsic standoff ratio, relaxation oscillator, 2N2646, base 1 base 2, SCR trigger, programmable unijunction transistor
A unijunction transistor, or UJT, is a three-terminal semiconductor device with only one pn junction, built on a bar of lightly doped silicon that has an ohmic contact at each end…
unilm unified language model, unilm unified, unified language, cnn glue mask mt nlg nlp, unilm unified language, unified language model, language model
UniLM - Unified Language Model Introduction & Motivation UniLM: unified pre-training for diverse NLP tasks via shared architecture. Supports understanding, generation, machine…
union-find disjoint sets, union-find disjoint, disjoint sets
Union-Find Disjoint Sets Introduction & Motivation Union-Find Disjoint Sets addresses a central problem in data-structures process control and operation-amortized-nearly-constant…
generative models
UniPC sampling is the unified predictor-corrector sampling framework that achieves high-order diffusion integration with broad model compatibility - it is designed to deliver…
unipolar field-effect transistor, majority carrier device, Shockley 1952 unipolar, unipolar vs bipolar, JFET and MOSFET, single-carrier transistor
A unipolar transistor is a transistor in which current is carried by only one type of charge carrier, electrons or holes, flowing through a single channel of the same doping type…
Dimensional Analysis and Unit Conversion Validation Introduction & Motivation Dimensional Analysis and Unit Conversion Validation addresses a central problem in {subject}…
interpretability
Universal Adversarial is input-agnostic perturbations that cause misclassification across many different samples - It demonstrates shared vulnerabilities in learned…
Universal Adversarial Perturbations Introduction & Motivation Universal Adversarial Perturbations addresses a central problem in robustness of advrobust models under adversarial…
Universal Adversarial Perturbations (Extended Study) Introduction & Motivation Universal Adversarial Perturbations (Extended Study) addresses a central problem in robustness…
Universal Adversarial Perturbations (Extended Study) Introduction & Motivation Universal Adversarial Perturbations (Extended Study) addresses a central problem in robustness…
ai safety
Universal adversarial triggers are short sequences of tokens that, when prepended or appended to any input, reliably cause a language model to produce specific unwanted behaviors…
Cybenko 1989, Hornik 1989, Stinchcombe and White, Funahashi 1989, Leshno 1993, width and depth, expressive power, universal approximator
The universal approximation theorem states that a feedforward neural network with a single hidden layer containing enough units and a suitable nonlinear activation function can…
standards
UCIe (Universal Chiplet Interconnect Express) is an open industry standard for connecting chiplets — separate silicon dies — together inside a single package. As monolithic chips…
universal chiplet interconnect protocol link training prediction, universal chiplet interconnect, chiplet interconnect protocol, universal chiplet
Universal Chiplet Interconnect Protocol Link Training Prediction with Machine Learning Introduction & Motivation Universal Chiplet Interconnect Protocol Link Training Prediction…
universal chiplet interconnect test yield prediction, universal chiplet interconnect, chiplet interconnect test, interconnect test yield, test yield prediction
Universal Chiplet Interconnect Test Yield Prediction with Machine Learning Introduction & Motivation Universal Chiplet Interconnect Test Yield Prediction with Machine Learning…
universal Turing machine, program as data, stored program, description number, virtual machine, interpreter, 1936 universal machine
The universal computing machine, now usually called the universal Turing machine, is the single machine Alan Turing described in section 6 of his 1936 paper that can do the work…
domain adaptation
Universal Domain Adaptation (UniDA) is a domain adaptation setting where the source and target domains may have different label sets—with categories that are private to…
Universal Domain Adaptation for Unknown Label Sets Introduction & Motivation Universal Domain Adaptation for Unknown Label Sets addresses a central problem in transfer learning…
Universal Domain Adaptation Introduction & Motivation Universal Domain Adaptation addresses a central problem in transfer learning and {target} adaptation: how to Apply {subject}…
llm architecture
Universal Transformers are a generalization of the standard transformer architecture that applies the same transformer layer (with shared weights) repeatedly to the input sequence…
reinforcement learning advanced
Universal Value Function is value-function approximation that generalizes across states and goals in one model. - It predicts expected return for arbitrary goal conditions instead…
uvfa, reinforcement learning
UVFA (Universal Value Function Approximators) is a framework for generalizing value functions across goals — extending standard value functions to that estimate the expected…
neural architecture
Universally Slimmable Networks (US-Nets) are an extension of slimmable networks that support any arbitrary width multiplier, not just preset values — enabling continuous…
universities, academic, research, student, professor, education
Yes, we actively support universities and research institutions through our Academic Program offering 50% discounts on MPW services, free design tool training, and technical…
audio & speech
UnivNet is a universal GAN-based neural vocoder designed for multi-speaker and multi-domain audio synthesis. - It targets strong waveform quality without per-speaker fine-tuning…
unk, nlp
UNK token is the special placeholder token used when input text contains symbols or words not represented by the tokenizer vocabulary - it provides fallback handling…
forget, remove
Machine Unlearning What is Machine Unlearning? Removing specific knowledge, behaviors, or data influence from a trained model without full retraining. Why Unlearning? Reason…
ai safety
Unlearning removes specific knowledge or capabilities from trained models for safety, privacy, or compliance. Motivations: Remove copyrighted content, forget personal data (GDPR…
time series models
Unobserved components is latent time-series components such as trend and cycle that are inferred from observed signals - State-space estimation recovers hidden components…
Rosenbaum Bounds for Sensitivity to Unmeasured Confounders Introduction & Motivation Rosenbaum Bounds for Sensitivity to Unmeasured Confounders addresses a central problem…
bare wafer, substrate inspection, particle detection, surface defect, metrology, substrate
Unpatterned wafer inspection is the metrology process of examining bare silicon wafers before any patterning — using optical, laser scattering, or surface scanning techniques…
unpatterned wafer particle count drift prediction machine learning, unpatterned wafer, particle count drift, unpatterned wafer particle count drift prediction
Unpatterned Wafer Particle Count Drift Prediction with Machine Learning Introduction & Motivation Unpatterned Wafer Particle Count Drift Prediction with Machine Learning addresses…
manufacturing operations
Unplanned Downtime is unexpected equipment stoppage that interrupts production outside scheduled events - It is a major source of availability loss and schedule instability. What…
emergency repair, equipment breakdown
Unplanned Maintenance refers to emergency equipment repairs triggered by unexpected failures, as opposed to scheduled preventive maintenance. What Is Unplanned Maintenance?…
Unrolled GAN Training Introduction & Motivation Unrolled GAN Training addresses a central problem in generative generation and mode-collapse-prevention synthesis: how to Design…