ScholarGate
Khám phá
Thư việnThư viện của tôiBànKiểm tra trướcReview StudioTrợ lý
Không gian làm việc
So sánh
Xây dựng kệ sách của bạn

Lưu phương pháp, sắp xếp bộ sưu tập và mang chúng đến bàn của bạn.

Tạo tài khoản
Thư viện
 / Duyệt
Đăng nhập
Thư viện

Khám phá khoa học theo phương pháp, lĩnh vực và bằng chứng.

Một danh mục duy nhất về các phương pháp nghiên cứu — tìm hiểu cách mỗi phương pháp hoạt động, khi nào nên dùng và điều nó không làm được.

6,435 phương pháp11 lĩnh vực7 họ phương pháp40 ngôn ngữ
Atlas khoa họcLập bản đồ cấu trúc của khoa học trước khi sử dụng.Lĩnh vực · phương pháp · lộ trình bằng chứngKhám phá bản đồ
Lĩnh vựcHealth & Medicine716Psychology570Business & Finance410Engineering330Life Sciences263Education261Research Practice
ScholarGate

Thư viện tham khảo về phương pháp nghiên cứu, đặt nội dung lên hàng đầu — mỗi phương pháp là gì, hoạt động ra sao và bắt nguồn từ đâu.

Dữ liệu mở (CC-BY)

Khám phá

  • Thư viện
  • Tìm phương pháp…
  • Duyệt theo lĩnh vực
  • Lĩnh vực
  • Hành trình
  • So sánh
  • Nên dùng phương pháp nào?

Tham khảo

  • Lĩnh vực
  • Bản đồ
  • Bảng thuật ngữ
  • Phương pháp luận
  • Triết học

Không gian làm việc

  • Thư viện của tôi
  • Bàn
  • Trò chuyện

Công ty

  • Giới thiệu
  • Bảng giá
  • Liên hệ
  • Đề xuất phương pháp

Các mục từ được biên soạn từ những nguồn đã công bố nhằm mục đích tham khảo. Việc kiểm chứng tính chính xác và mức độ phù hợp của bất kỳ thông tin nào cho mục đích sử dụng của bạn vẫn thuộc trách nhiệm của bạn.

© 2026 ScholarGate · Thư viện tham khảo về phương pháp nghiên cứu
  • Quyền riêng tư
  • Cookie
  • Điều khoản
  • Xóa tài khoản
248
Natural Sciences236
Social Sciences185
Environment & Sustainability160
Law30
Phương phápThống kê1,836Trí tuệ nhân tạo & học máy1,661Khoa học quyết định932Phương pháp nghiên cứu1,354Đo lường1,745Nhân quả & bằng chứng532Thực hành nghiên cứu118
1,522 phương pháp · Trí tuệ nhân tạo & học máyXóa
Các phương pháp thực khớp với bộ lọc của bạn.
Sắp xếpĐộ phổ biếnA–ZZ–AMới nhất
mining engineering

Pseudoflow

The Pseudoflow Algorithm, developed by Dorit Hochbaum in 1992, is a polynomial-time algorithm for computing maximum weighted closures in directed acyclic graphs. In mining, it solves the ultimate pit limit problem more efficiently than earlier methods. By maintaining feasible pseudoflows and iteratively eliminating neg

2 nguồn1992
acoustics

Psychoacoustic Masking

Psychoacoustic masking describes how the human auditory system suppresses the perception of weak sounds in the presence of stronger sounds. Formalized by Eberhard Zwicker in the 1960s, masking is a fundamental phenomenon in hearing and the basis for perceptual audio coding (MP3, AAC, OPUS). Masking occurs both in frequ

3 nguồn1961
linguistics

Psycholinguistic Eye-Tracking

Psycholinguistic Eye-Tracking is a method that measures eye movements during reading or visual processing to investigate how the mind processes language. Pioneered by Keith Rayner, eye-tracking reveals which parts of text attract attention, how long readers spend on different words, and how eye movements relate to comp

3 nguồn1975
thermodynamics

Psychrometric Analysis

Psychrometric analysis is the study of humid air (air-water vapor mixtures) and its properties. It is essential for designing and analyzing air conditioning, ventilation, and dehumidification systems. Psychrometric analysis relates dry-bulb temperature, wet-bulb temperature, dew point, relative humidity, and specific h

2 nguồn1911
astronomy

Pulsar Timing Array

A pulsar timing array uses multiple millisecond pulsars as a distributed network of gravitational wave detectors across the galaxy. Proposed theoretically by Stephen Detweiler in 1979, this method exploits the extraordinary timing precision of pulsars to detect the subtle spacetime distortions caused by gravitational w

3 nguồn1979
civil engineering

Pushover Analysis

Pushover analysis is a nonlinear static method for assessing seismic structural performance. Introduced by Fajfar in 1996 as part of the N2 method, it progressively increases lateral loads on a structure until it reaches a target displacement, revealing how structures deform and yield under seismic events.

3 nguồn1996
deep learning

Pyraformer

Pyraformer is a Transformer-based model for long-range time-series forecasting introduced by Liu et al. at ICLR 2022. Its central innovation is a Pyramidal Attention Module (PAM) that organizes tokens into a multi-resolution hierarchy, enabling the model to capture temporal dependencies across multiple scales while kee

1 nguồn2022
machine learning

Q-Learning

Q-learning, introduced by Christopher Watkins and Peter Dayan in 1992, is a model-free reinforcement-learning algorithm that learns the value of taking each action in each state — the Q-function — purely from experience, without a model of the environment. It is off-policy: it learns the optimal action-values while fol

2 nguồn1992
mining engineering

Q-System

The Q-System (NGI Index), introduced by Nick Barton and colleagues at the Norwegian Geotechnical Institute in 1974, is an alternative rock mass classification to RMR. It combines six parameters into a dimensionless index Q ranging from 0.001 to 1000, where higher Q values indicate better rock quality. The Q-System is p

2 nguồn1974
deep learning

QLoRA

QLoRA is an efficient fine-tuning method introduced by Dettmers et al. in 2023 that enables fine-tuning large language models using quantization and low-rank adaptation. By combining 4-bit quantization with LoRA, QLoRA reduces memory requirements by 75%, enabling fine-tuning of 65B-parameter models on single GPUs.

1 nguồn2023
bioinformatics

QSAR

Quantitative Structure-Activity Relationship (QSAR) modeling predicts biological activity from molecular structure using statistical or machine learning models. Pioneered by Hansch in 1964, QSAR correlates numerical molecular descriptors with measured bioactivity, enabling prediction of activity for untested compounds

3 nguồn1964
genetics

QTL Mapping

Quantitative trait loci (QTL) mapping is a genetic method that localizes chromosomal regions influencing quantitative traits—continuous phenotypes controlled by multiple genes and environmental factors. Developed by Lander and Botstein in 1989, QTL mapping uses linkage analysis and trait variation in segregating popula

3 nguồn1989
machine learning

Quadratic Discriminant Analysis

Quadratic discriminant analysis is a generative classifier that models each class with its own multivariate Gaussian distribution, allowing each class a separate covariance matrix. Unlike linear discriminant analysis, which assumes a shared covariance and yields linear boundaries, QDA's per-class covariances produce cu

2 nguồn1939
quantum computing

Quantum Approximate Optimization Algorithm

The Quantum Approximate Optimization Algorithm (QAOA) is a hybrid quantum-classical algorithm designed to solve combinatorial optimization problems on near-term quantum devices. Introduced by Farhi, Goldstone, and Gutmann in 2014, QAOA encodes optimization problems into quantum circuits and uses classical optimization

3 nguồn2014
quantum computing

Quantum Key Distribution (BB84)

Quantum Key Distribution (QKD) BB84 is a cryptographic protocol allowing two parties to establish a shared secret key using quantum mechanics. Proposed by Bennett and Brassard in 1984, BB84 provides information-theoretic security: an eavesdropper's presence is guaranteed to be detected, and the secret key is provably s

3 nguồn1984
quantum computing

Quantum Monte Carlo

Quantum Monte Carlo (QMC) is a stochastic computational method for computing ground state properties of quantum many-body systems. Combining classical Monte Carlo sampling with quantum mechanics, QMC approaches are among the most accurate methods available for electronic structure and condensed matter physics, achievin

3 nguồn1953
quantum computing

Quantum Phase Estimation

Quantum Phase Estimation (QPE) is a fundamental quantum subroutine that estimates the eigenvalues of a unitary operator. Developed by Alexei Kitaev in 1995, QPE combines controlled unitary evolution with the quantum Fourier transform to extract eigenvalues from quantum states with exponential precision scaling.

3 nguồn1995
quantum computing

Quantum SVM

Quantum Support Vector Machine (QSVM) is a quantum machine learning algorithm combining quantum feature spaces with classical SVM training. Proposed by Rebentrost et al. in 2014, QSVM leverages quantum processors to compute kernel functions, potentially offering speedup for classification problems while remaining pract

3 nguồn2014
quantum computing

Quantum Teleportation

Quantum Teleportation is a protocol for transferring an unknown quantum state between distant parties using entanglement and classical communication. Discovered by Bennett et al. in 1993, teleportation violates no fundamental principles but demonstrates the power of entanglement: an unknown quantum state can be reconst

3 nguồn1993
meteorology

Quasi-Geostrophic Omega Equation

The quasi-geostrophic (QG) omega equation is a fundamental diagnostic equation in synoptic meteorology that relates vertical motion (omega = dP/dt) to horizontal temperature and vorticity fields. It predicts where air rises and sinks based on the geostrophic flow structure without explicitly solving for vertical veloci

2 nguồn1970
aerospace

Quaternion Attitude

Quaternion attitude representation is a mathematical framework for describing three-dimensional rotations using four-dimensional vectors (quaternions). Superior to Euler angles due to the absence of singularities (gimbal lock), quaternions are the standard representation in modern attitude estimation, spacecraft contro

3 nguồn1843
text mining

Question Answering

Question answering is a natural-language-processing task that automatically answers natural-language questions grounded in a given context passage, using either extractive or generative approaches. The task was crystallised by the SQuAD benchmark of Rajpurkar et al. (2016), and later models such as XLNet (Yang et al.,

2 nguồn
model evaluation

R-squared

The coefficient of determination, denoted R², measures the proportion of variance in the dependent variable explained by the independent variables in a regression model. Introduced by Karl Pearson in the late 19th century, R² is one of the most widely used metrics for assessing how well a model fits observed data.

3 nguồn1896
applied physics

Radial Velocity Method

The radial velocity method detects exoplanets by measuring the Doppler shift of a star's spectral lines caused by gravitational tugging from orbiting planets. When a planet orbits a star, the star wobbles slightly toward and away from Earth, creating periodic shifts in its light spectrum. First proposed by Friedrich Wi

3 nguồn1844
nuclear physics

Radiation Dose Assessment

Radiation dose assessment is a systematic evaluation of human exposure to ionizing radiation from external or internal sources, formalized by the International Commission on Radiological Protection (ICRP) in the late 20th century. It combines radiation transport calculations with biological effect models to quantify ab

2 nguồn1928
nuclear physics

Radiation Protection Optimization

Radiation protection optimization is a systematic approach to design and manage exposure reduction strategies using risk-benefit analysis, codified by the ICRP in the principle of As Low As Reasonably Achievable (ALARA) in 1977. By balancing radiation dose reduction against cost, effort, and societal benefit, it guides

2 nguồn1977
nuclear physics

Radiation Shielding Design

Radiation shielding design is an engineering discipline that uses physics-based calculations and materials selection to reduce radiation exposure to acceptable levels, originating from Curie and Rutherford's early radiation studies in the 1890s. By combining attenuation theory, source characterization, and dose modelin

2 nguồn1898
astronomy

Radiative Transfer

Radiative transfer is the mathematical treatment of how light propagates through matter, including absorption, emission, and scattering. Central to astrophysics and stellar atmosphere modeling, radiative transfer calculations translate physical conditions (density, temperature, composition) into observable spectra and

3 nguồn1978
nuclear physics

Radioactive Waste Classification

Radioactive waste classification is a systematic framework for categorizing radioactive materials based on activity, heat generation, and long-term hazard potential, developed by the IAEA. It stratifies waste into classes (exempt, very low-level, low-level, intermediate-level, high-level) to determine appropriate manag

2 nguồn1960
geophysics

Radiocarbon Dating

Radiocarbon dating is a radiometric technique that determines the age of organic materials by measuring the radioactive decay of ¹⁴C (carbon-14), a rare isotope produced in the atmosphere by cosmic ray interactions. Developed by Willard Libby in 1949, radiocarbon dating became a foundational method in archaeology, pale

2 nguồn1949
reliability engineering

Rainflow Counting

Rainflow counting is a fatigue cycle counting method that converts a complex stress history into individual cycles for damage assessment. Developed by Tatsuo Endo and colleagues in 1974, it provides the most physically realistic representation of fatigue damage when combined with Miner's linear cumulative damage hypoth

4 nguồn1974
materials science

Raman Deconvolution

Raman Deconvolution is the mathematical decomposition of experimental Raman spectra into constituent peaks using spectral fitting algorithms. Building on Raman spectroscopy (discovered by C.V. Raman in 1928), Raman deconvolution resolves overlapping vibrational bands into individual component peaks, revealing detailed

3 nguồn1928
machine learning

Random Forest

Random Forest is an ensemble learning method, introduced by Leo Breiman in 2001, that grows many decision trees on bootstrap samples of the data and combines their votes to produce strong classification and regression. By pooling many slightly different trees, it produces more accurate and more stable predictions than

2 nguồnadvanced2001
machine learning

Random Projection

Random projection reduces dimensionality by multiplying the data by a random matrix, relying on the Johnson-Lindenstrauss lemma (1984), which guarantees that projecting onto enough random directions approximately preserves all pairwise distances. Unlike PCA it does not analyze the data at all — the projection is random

2 nguồn1984
thermodynamics

Rankine Cycle

The Rankine Cycle is the fundamental thermodynamic cycle for steam power plants. It describes how thermal energy from burning fuel or concentrated solar radiation is converted to mechanical work and ultimately electricity. The cycle consists of four processes: isobaric heat addition in the boiler, isentropic expansion

2 nguồn1859
control theory

Rapidly-Exploring Random Tree

The Rapidly-Exploring Random Tree (RRT) is a motion planning algorithm that builds a tree of feasible paths by iteratively sampling random configurations in the workspace and connecting them to the nearest existing node in the tree. Introduced by LaValle in 1998, RRT is a breakthrough for high-dimensional motion planni

3 nguồn1998
telecommunications

Ray Tracing Propagation

Ray tracing is a deterministic propagation modeling technique for predicting electromagnetic field strength at specific locations. Instead of empirical formulas (like Okumura-Hata), ray tracing traces paths of electromagnetic energy as it reflects, diffracts, and scatters off buildings and terrain. With accurate 3D geo

2 nguồn1993
applied physics

Reactive Distillation

Reactive distillation couples reaction and separation in a single column, where reactants are separated from products continuously while simultaneously undergoing reaction on catalytic trays. Pioneered in the 1990s by Klaus Sundmacher and others, this process intensification technique dramatically reduces capital cost,

3 nguồn1995
electrical engineering

Reactive Power Compensation

Reactive power compensation adjusts the flow of reactive power (VARs) in electrical networks to support voltage profiles, reduce losses, and improve power factor. Methods include fixed capacitor banks, switched capacitors, synchronous condensers, and FACTS devices. Proper compensation is essential for maintaining volta

3 nguồn1920
nuclear physics

Reactor Kinetics

Reactor kinetics is the study of neutron population dynamics in a reactor core, originating from Fermi's first controlled chain reaction in 1942. It models power changes in response to control rod movements, temperature feedback, and accidental transients using coupled differential equations accounting for prompt and d

2 nguồn1942
text mining

Readability Analysis

Readability analysis measures how well a text suits its intended audience by applying established readability formulas such as Flesch-Kincaid and Gunning Fog. The modern formula family was derived by Kincaid and colleagues in 1975, and it turns prose into a single score or target reading-grade level that signals how ea

2 nguồn1975
model evaluation

Recall (Sensitivity)

Recall measures the proportion of actual positive cases that were correctly identified by the classifier. It answers the question: 'Of all the cases that were truly positive, how many did we find?' Recall is critical in scenarios where missing positive cases is costly.

2 nguồn
complex systems

Recurrence Quantification Analysis

Recurrence Quantification Analysis (RQA) is a nonlinear method for characterizing the dynamics of a time series by quantifying the small-scale structure of its recurrence plot. Introduced in its modern, comprehensive form by Marwan, Romano, Thiel, and Kurths in 2007, RQA extracts scalar measures — such as recurrence ra

1 nguồn2007
deep learning

Recurrent Neural Network

A Recurrent Neural Network (RNN) is a class of neural network designed to process sequential data by maintaining a hidden state that carries information across time steps. Introduced in its modern form by Rumelhart et al. (1986) and further shaped by Elman (1990), RNNs became the dominant architecture for sequence mode

2 nguồn1986
deep learning

Reformer

The Reformer is an efficient variant of the Transformer architecture introduced by Kitaev, Kaiser, and Levskaya at ICLR 2020. It addresses the prohibitive O(L²) memory and computational cost of standard self-attention for long sequences. The key innovations are locality-sensitive hashing (LSH) attention, which approxim

1 nguồn2020
machine learning

Regression Splines

Regression splines model a nonlinear relationship by fitting piecewise polynomials that join smoothly at a set of points called knots. Cubic and natural splines are the most common, and smoothing splines add a roughness penalty that automatically balances fit against smoothness. Splines are the standard flexible buildi

2 nguồn1996
machine learning

Regularized Boosting

Regularized boosting extends gradient boosting by adding explicit controls — shrinkage (learning rate), L1/L2 weight penalties, subsampling, and tree-complexity limits — to the objective function and the update rule. These constraints reduce overfitting, stabilise the model on noisy or small datasets, and are the core

2 nguồn2001
machine learning

Regularized CatBoost

Regularized CatBoost applies explicit regularization controls — L2 leaf regularization, tree depth constraints, shrinkage rate, and model size penalties — on top of CatBoost's ordered gradient boosting framework, reducing overfitting while retaining CatBoost's native handling of categorical features and its low predict

2 nguồn2018
machine learning

Regularized Decision Tree

A regularized decision tree is a decision tree model whose complexity is intentionally limited through pruning, depth constraints, or penalty terms to prevent overfitting. Rooted in Breiman et al.'s CART framework (1984), regularization converts the greedy tree-growing procedure into a bias-variance tradeoff, yielding

2 nguồn1984
machine learning

Regularized Federated Learning

Regularized federated learning extends the federated learning framework by adding penalty terms to each client's local objective, anchoring local updates closer to the global model. The canonical formulation — FedProx — adds a proximal term that controls how far any single client can drift, improving convergence and st

2 nguồn2020
machine learning

Regularized Few-Shot Learning

Regularized few-shot learning augments standard few-shot learning pipelines with explicit regularization mechanisms — such as weight decay, dropout, data augmentation, label smoothing, or manifold constraints — to reduce overfitting to the tiny support sets that define each episode. This produces more generalizable mod

2 nguồn2016
machine learning

Regularized Gaussian Mixture Model

A Regularized Gaussian Mixture Model (GMM) adds a small positive constant to the diagonal of each component covariance matrix during the Expectation-Maximization algorithm, preventing singular or near-singular matrices that cause numerical failures when the data are sparse, high-dimensional, or contain near-duplicate o

2 nguồn2000
machine learning

Regularized Gaussian Process

A Regularized Gaussian Process (GP) is a probabilistic kernel-based model that places a prior over functions and explicitly controls overfitting through a noise regularization parameter — the observation noise variance — that prevents the model from memorizing training labels. It produces calibrated uncertainty estimat

2 nguồn2006
machine learning

Regularized Gradient Boosting

Regularized gradient boosting extends the classic additive tree ensemble (Friedman 2001) by embedding L1 and L2 penalty terms directly into the training objective, along with a complexity penalty on tree size. Popularized by XGBoost (Chen & Guestrin 2016), this framework reduces overfitting and improves generalization

2 nguồn2001
machine learning

Regularized k-means

Regularized k-means extends standard k-means by adding a penalty term — most commonly an L1 (lasso-type) or L2 constraint — to the objective function. This discourages degenerate cluster solutions and, in the sparse variant introduced by Witten and Tibshirani (2010), simultaneously selects the features that drive clust

2 nguồn2010
machine learning

Regularized k-nearest neighbors

Regularized k-Nearest Neighbors (kNN) extends the classical nearest-neighbor algorithm by incorporating regularization mechanisms — most commonly kernel-based distance weighting or bandwidth control — that smooth predictions, reduce sensitivity to the choice of k, and lower variance. The result is a more stable and bet

2 nguồn1967
machine learning

Regularized Naive Bayes

Regularized Naive Bayes augments the classical Naive Bayes probabilistic classifier with explicit smoothing or shrinkage — most commonly Laplace (additive) smoothing — to prevent zero-probability estimates for unseen feature values and to reduce overfitting. The result is a fast, robust classifier that generalizes bett

2 nguồn1950
machine learning

Regularized Online Learning

Regularized online learning extends the online learning paradigm by incorporating a regularization penalty into each weight update, controlling model complexity while processing data one example at a time. Algorithms such as Follow-the-Regularized-Leader (FTRL) and Regularized Dual Averaging (RDA) make this approach pr

2 nguồn2007
machine learning

Regularized random forest

Regularized Random Forest (RRF), introduced by Deng and Runger in 2012, extends the standard Random Forest by adding a penalty that discourages splits on features not already used in the ensemble. This built-in regularization produces sparser, less redundant feature subsets, making the model especially valuable when fe

2 nguồn2012
machine learning

Regularized semi-supervised learning

Regularized semi-supervised learning adds explicit geometric or graph-based penalty terms to a semi-supervised objective so that the decision function varies smoothly over the data manifold. Pioneered through manifold regularization (Belkin, Niyogi & Sindhwani, 2006), it exploits the structure of both labeled and unlab

2 nguồn2006
← 1617 / 2618 →