按方法、领域与证据探索科学。

Counterfactual Explanations

Counterfactual explanations, introduced by Wachter, Mittelstadt, and Russell in 2017, answer the question: 'What is the smallest change to the input that would have produced a different model output?' Rather than explaining why a model made a decision, they describe what would need to change for that decision to be rev

quantum computing

Coupled Cluster CCSD

Coupled Cluster theory, particularly CCSD (Singles and Doubles) and CCSD(T) with perturbative triples, is one of the most accurate methods for molecular electronic structure. Developed by Jiri Cizek in 1966, CC theory treats the ground state wave function as an exponential of excitation operators applied to the Hartree

particle physics

CP Violation Measurement

Charge-Parity (CP) violation measurement is the experimental study of asymmetries between particle and antiparticle processes, a fundamental probe of physics beyond the Standard Model. By comparing decay rates and asymmetries in kaons, B mesons, and neutrinos, physicists constrain new sources of CP violation and addres

quantitative finance

Crank-Nicolson Pricing

The Crank-Nicolson method is a widely-used implicit finite difference scheme for solving PDEs in option pricing. It provides second-order accuracy in both space and time, unconditional stability, and can efficiently price derivatives with early exercise features (American options) or complex boundary conditions.

bioinformatics

CRISPR Screen Analysis

CRISPR screen analysis processes data from pooled genetic screens using CRISPR-Cas9 to identify genes required for cell growth, survival, or phenotype in specific conditions. Developed by Zhang, Sanjana, and others, this computational pipeline transforms sequencing readouts of guide RNA abundances into ranked lists of

text mining

Cross-Document Entity Tracking

Cross-document entity tracking, formally known as cross-document coreference resolution, identifies and merges all references to the same real-world entity scattered across a collection of documents. Rooted in the B3 evaluation framework introduced by Bagga and Baldwin (1998) and substantially advanced by the neural jo

text mining

Cross-lingual Text Analysis

Cross-lingual text analysis lets you compare and analyse texts written in different languages within a shared vector space. Building on multilingual representation learning surveyed by Conneau et al. (2020) and Pires et al. (2019), it maps documents from several languages into one common embedding space so multilingual

time series

Cross-Wavelet Transform

The cross-wavelet transform (XWT) is a bivariate extension of the continuous wavelet transform that measures the joint time-frequency representation of two signals. Introduced by Torrence and Compo (1998) and applied extensively by Grinsted, Moore, and Jevrejeva (2004) to geophysical data, XWT reveals where two signals

deep learning

Crossformer

Crossformer is a Transformer-based architecture for multivariate time series forecasting, introduced by Yunhao Zhang and Junchi Yan at ICLR 2023. Unlike earlier Transformer variants that treat each variate independently, Crossformer explicitly models cross-dimension dependencies alongside temporal patterns. It achieves

bioinformatics

Cryo-EM Reconstruction

Cryo-electron microscopy (cryo-EM) determines three-dimensional macromolecular structures at atomic or near-atomic resolution by imaging proteins frozen in vitreous ice. Pioneered by Frank, Henderson, and others, this technique has revolutionized structural biology by enabling visualization of large, non-crystallizable

telecommunications

CSMA/CA

CSMA/CA is a random access protocol for wireless medium access control, designed to enable multiple devices to share a wireless channel while minimizing collisions. Introduced by Phil Karn in 1990, it is the foundation of WiFi (IEEE 802.11) and is now the de facto standard for unlicensed spectrum access. CSMA/CA combin

applied physics

CSTR Model

The CSTR (Continuous Stirred-Tank Reactor) model describes the behavior of an ideal mixed reactor where fresh feed is continuously added, products are withdrawn, and contents are kept uniform by vigorous stirring. This fundamental model, formalized by Octave Levenspiel in the 1960s, is widely used to design and scale b

oceanography

CTD Profiling

Conductivity-Temperature-Depth (CTD) profiling is the primary method for measuring vertical profiles of seawater properties in oceanography. Developed by Neil Brown in 1977, CTD instruments are equipped with sensors for conductivity, temperature, and pressure (depth), and are typically mounted on water-sampling rosette

deep learning

Curriculum Learning

Curriculum Learning is a training strategy for machine learning models, introduced by Bengio et al. in 2009, in which training examples are presented in a meaningful order—typically from easy to hard—rather than at random. Inspired by how humans and animals learn progressively, it organizes training data into a curricu

numerical methods

Cyclomatic Complexity

Cyclomatic Complexity (CC), introduced by Thomas McCabe in 1976, is a quantitative metric measuring the number of linearly independent paths through a function's control-flow graph. A function with high cyclomatic complexity is harder to understand, test, and maintain; McCabe advocated a threshold of 10 as the complexi

deep learning

Data Augmentation

Data augmentation is a family of techniques that artificially expands a training dataset by applying label-preserving transformations to existing samples. Originally systematized for image classification tasks, it is now applied broadly across vision, text, audio, and tabular domains. It emerged as a practical answer t

data fusion

Data Fusion

Data fusion is a multi-level process that combines data and information from multiple sensors and sources to achieve improved accuracy, completeness, and confidence in estimates that cannot be obtained from any single source alone. Formally introduced as the Joint Directors of Laboratories (JDL) model by Hall and Llina

model evaluation

Davies-Bouldin Index

The Davies-Bouldin Index, introduced by Davies and Bouldin in 1979, is a metric for evaluating clustering quality based on the average similarity between each cluster and its most similar neighboring cluster. Lower values indicate better clustering, with a minimum of 0 representing perfectly separated, non-overlapping

architecture

Daylight Simulation

Daylight Simulation is a computational method for predicting the availability and distribution of daylight in interior spaces and assessing visual comfort under varying sky conditions. Developed by researchers like Christoph Reinhart and John Mardaljevic in the 2000s, it has become central to designing healthy, energy-

machine learning

DBSCAN

DBSCAN is a density-based clustering algorithm, introduced by Ester, Kriegel, Sander and Xu in 1996, that groups together points lying in dense regions and flags points in sparse regions as noise. It is effective on noisy data and on clusters of irregular, non-spherical shapes.

bioinformatics

De Novo Transcriptome Assembly

De novo transcriptome assembly reconstructs full-length messenger RNA sequences directly from sequencing reads without requiring a reference genome. Pioneered by Regev, Haas, and colleagues, this pipeline enables transcript discovery in non-model organisms and detection of novel isoforms, fusion genes, and splice varia

aerospace

Dead Reckoning

Dead Reckoning is a fundamental navigation method that estimates position and heading by integrating velocity and angular rate measurements from inertial sensors over time, without external references such as GPS. The term derives from maritime tradition ('deduced reckoning') and remains a cornerstone of aerospace and

machine learning

Decision Tree

A Decision Tree is an interpretable classification and regression method, formalised by Breiman, Friedman, Olshen and Stone in their 1984 CART framework, that partitions the data with hierarchical if-then rules. Each split sends observations down one branch or another until a prediction is read off the leaf.

deep learning

Deep Belief Network

A Deep Belief Network is a generative probabilistic model composed of multiple layers of stochastic, latent variables. Introduced by Hinton, Osindero, and Teh in 2006, DBNs were among the first deep architectures to be trained efficiently. Each pair of adjacent layers forms a Restricted Boltzmann Machine, and the netwo

cryptography

Deep Packet Inspection

Deep Packet Inspection (DPI) is a network traffic analysis technique that examines the complete packet payload beyond header information to identify, classify, and potentially control data traffic. Developed in the 1990s for network monitoring and management, DPI analyzes packet contents to detect protocols, applicatio

deep learning

Deep Reinforcement Learning

Deep Reinforcement Learning combines neural networks with reinforcement learning so an agent learns by interacting with an environment, popularised by Mnih and colleagues' 2015 Nature work on human-level Atari control. Instead of learning from a fixed labelled dataset, the agent takes actions, observes rewards, and gra

deep learning

DeepAR

DeepAR is Amazon's industrial forecasting model, introduced by Salinas, Flunkert and Gasthaus (2017; published 2020), that uses an autoregressive recurrent neural network to estimate the parameters of a probability distribution at each step, producing a confidence interval rather than a single point forecast. It can mo

software engineering

Defect Prediction Model

Defect prediction models forecast the likelihood of software faults in code modules using statistical or machine learning approaches. Pioneered by Ostrand, Weyuker, and Bell (2005), these models correlate code metrics (complexity, churn, coupling) with historical defect data to identify high-risk components. Organizati

network analysis

Degree Centrality

Degree centrality is the simplest and most intuitive measure of a node's importance in a network, defined as the number of direct ties a node has to other nodes. Normalized by dividing by the maximum possible ties, it allows comparison across networks of different sizes and is the starting point of almost every network

oceanography

Degree Heating Weeks

Degree Heating Weeks (DHW) is a thermal stress metric that quantifies accumulated heat exposure above a coral bleaching threshold, computed from satellite sea surface temperature data. Developed by NOAA's Coral Reef Watch program in 2003, DHW provides a standardized index for predicting and monitoring coral bleaching s

ensemble learning

Dempster-Shafer Fusion

Dempster-Shafer fusion is an ensemble method based on evidence theory (belief functions) that combines predictions from multiple sources by assigning basic probability masses to subsets of hypotheses. Rather than requiring a probability distribution over single outcomes, it allows uncertainty over sets of outcomes, pro

manufacturing

Denavit-Hartenberg Parameters

The Denavit-Hartenberg (DH) convention is a systematic mathematical method for assigning coordinate frames to the links of an articulated robot or mechanism, enabling compact representation and computation of forward and inverse kinematics. Introduced by Denavit and Hartenberg in 1955, this method uses only four parame

deep learning

DenseNet

DenseNet (Densely Connected Convolutional Network), introduced by Huang, Liu, van der Maaten, and Weinberger at CVPR 2017 (Best Paper Award), connects every layer to every subsequent layer within a dense block so that each layer receives the concatenated feature maps of all preceding layers — maximising feature reuse,

quantum computing

Density Functional Theory

Density Functional Theory (DFT) is a computational method for determining the properties of materials and molecules by modeling the ground state electron density. Developed by Walter Kohn and Lu Jeu Sham in the 1960s, DFT reduces the complexity of quantum chemistry from tracking individual electron coordinates to optim

text mining

Dependency Parsing

Dependency parsing is a natural-language-processing task that reveals the syntactic dependency relations between the words of a sentence as a tree structure. Surveyed in the dependency-grammar tradition by Nivre (2005) and made fast and accurate with neural networks by Chen and Manning (2014), it is commonly used as a

manufacturing

Design for Manufacturing and Assembly

Design for Manufacturing and Assembly (DFMA) is a systematic methodology for creating products that are inherently easier and less expensive to manufacture and assemble. Developed by Boothroyd, Dewhurst, and Knight, DFMA evaluates design choices based on their impact on production cost, quality, and speed, guiding desi

fluid dynamics

Detached Eddy Simulation

Detached Eddy Simulation (DES) is a hybrid turbulence modeling approach introduced by Spalart in 1997 that combines the computational efficiency of RANS in attached boundary layers with the accuracy of LES in separated wake regions. By automatically switching between RANS and LES based on local grid spacing and turbule

simulation

Deterministic Agent-Based Modeling

Deterministic Agent-Based Modeling (D-ABM) is a computational simulation approach in which autonomous agents follow fully specified, non-random behavioral rules within a structured environment. Every run with identical initial conditions produces identical outcomes, making the model fully reproducible and transparent f

simulation

Deterministic Cellular Automata

Deterministic Cellular Automata (DCA) is a simulation method that models the evolution of complex systems through a regular grid of cells, each holding a discrete state, updated synchronously at each time step according to a fixed, deterministic rule applied to the cell and its neighbors. The outcome is fully reproduci

simulation

Deterministic Discrete-Event Simulation

Deterministic Discrete-Event Simulation (Deterministic DES) models a system as a sequence of events occurring at precise, pre-specified times using fixed input parameters. Unlike stochastic DES, no probability distributions are sampled; every arrival, service time, and resource availability is known in advance, making

simulation

Deterministic Markov Model

A Deterministic Markov Model is a cohort-level state-transition model in which all transition probabilities, state utilities, and costs are assigned single fixed values and the model is solved analytically in a single pass. Widely used in health technology assessment, policy analysis, and operations research, it traces

simulation

Deterministic Microsimulation

Deterministic Microsimulation applies a fixed set of policy rules or behavioral equations to each individual or household record in a microdata file, computing exact outcomes without any random sampling. It is the standard engine behind tax-benefit calculators and demographic projection models used by governments world

simulation

Deterministic Multi-Objective Optimization

Deterministic Multi-Objective Optimization (Deterministic MOO) is a family of classical optimization approaches that simultaneously minimize or maximize multiple conflicting objective functions over a deterministic feasible set. It produces a Pareto front — the set of non-dominated solutions — from which a decision-mak

simulation

Deterministic Scenario Analysis

Deterministic Scenario Analysis (DSA) is a structured planning method in which analysts construct a finite set of internally consistent future scenarios, each defined by fixed, precisely specified parameter values rather than probability distributions. By running a model or calculation under each scenario's fixed input

simulation

Deterministic Sensitivity Analysis

Deterministic Sensitivity Analysis (DSA) tests how model outputs change when individual or combined input parameters are varied across plausible ranges, one at a time or in structured combinations, without invoking probabilistic sampling. It is the standard approach in economic modeling, decision trees, and mathematica

simulation

Deterministic System Dynamics

Deterministic System Dynamics is the classical form of System Dynamics introduced by Jay Forrester in 1961, using fixed (non-probabilistic) ordinary differential equations to simulate stock-and-flow structures and feedback loops over time. All model parameters and relationships are specified as single-valued constants

deep learning

DETR (Detection Transformer)

DETR (Detection Transformer) is an end-to-end framework for object detection introduced by Carion et al. in 2020 that reformulates detection as a direct set prediction problem using transformers. Unlike traditional approaches that use hand-crafted post-processing like non-maximum suppression, DETR treats object detecti

linguistics

Dialectometry

Dialectometry is a quantitative method for measuring linguistic distances between dialects or languages using objective metrics applied to phonological, lexical, or phonetic data. Pioneered by Jean Seguy in 1973, dialectometry compares word lists, pronunciations, or phonetic transcriptions across speech varieties to ca

text mining

Dialogue Act Classification

Dialogue act classification is a natural-language-processing task that automatically labels the communicative function of each utterance in a conversation — such as question, answer, greeting, or rejection. Consolidated by Jurafsky et al. (1997) and Stolcke et al. (2000), it is a foundational component for chatbots and

bioinformatics

Differential ChIP-seq peak calling

Differential ChIP-seq peak calling identifies genomic loci where a protein of interest — typically a transcription factor or histone mark — shows significantly altered binding or occupancy between two or more biological conditions. By combining standard ChIP-seq peak detection with count-based statistical testing, the

bioinformatics

Differential Copy Number Variation Analysis

Differential copy number variation (dCNV) analysis identifies genomic regions where DNA copy numbers differ systematically between two conditions — such as tumor versus normal tissue, case versus control cohorts, or treated versus untreated cells. By combining probe-level read-depth or array-intensity data with statist

cryptography

Differential Cryptanalysis

Differential cryptanalysis is a statistical attack technique on symmetric block ciphers that analyzes differences in inputs and outputs to recover secret keys. Introduced by Eli Biham and Adi Shamir in 1990, differential cryptanalysis was the first practical attack on DES that outperformed brute force search. The techn

bioinformatics

Differential Epigenome-Wide Association Study

A Differential Epigenome-Wide Association Study (Differential EWAS) scans hundreds of thousands of CpG methylation sites across the genome to identify those whose methylation levels differ significantly between two or more comparison groups — such as cases vs. controls, exposed vs. unexposed, or distinct developmental

bioinformatics

Differential eQTL Analysis

Differential eQTL analysis identifies genetic variants — expression quantitative trait loci — whose regulatory effect on gene expression varies systematically across biological conditions such as tissue types, disease states, developmental stages, or treatment groups. By testing for statistical interactions between gen

bioinformatics

Differential Metabolomics Analysis

Differential metabolomics analysis is a computational pipeline that identifies metabolites whose abundance levels differ significantly between two or more biological conditions — such as disease versus control, treated versus untreated, or different developmental stages. By integrating mass spectrometry or NMR data wit

bioinformatics

Differential pathway enrichment analysis

Differential pathway enrichment analysis identifies biological pathways whose enrichment signals differ significantly between two or more experimental conditions — for example, between two diseases, two treatments, or two cell types. Rather than asking which pathways are enriched in one condition, it asks which pathway

privacy

Differential Privacy

Differential privacy is a mathematical framework for releasing statistical information about a dataset while providing rigorous guarantees that individual records cannot be identified or inferred. Introduced by Cynthia Dwork in 2006, it formalizes privacy as a probabilistic bound: any single individual's presence or ab

bioinformatics

Differential proteomics analysis

Differential proteomics analysis is a quantitative pipeline that identifies proteins whose abundance levels change significantly between two or more biological conditions — such as healthy versus diseased tissue, treated versus untreated cells, or different developmental stages. By combining mass spectrometry-based det

materials science

Differential Scanning Calorimetry

Differential Scanning Calorimetry (DSC) is a thermal characterization technique that measures the heat flow required to maintain a sample and an inert reference at the same temperature while both are heated or cooled. Invented by Watson, O'Neill, and colleagues in 1964, DSC directly quantifies enthalpy changes during p

bioinformatics

Differential single-cell RNA-seq analysis

Differential single-cell RNA-seq (scRNA-seq) analysis is a computational pipeline that compares transcriptomic profiles across biological conditions — such as treated versus untreated, disease versus healthy, or time points — at single-cell resolution. It identifies which genes, cell types, and cell states change betwe