Fact-checked by Grok 2 weeks ago

Robust statistics

Robust statistics is a branch of statistics focused on developing methods that remain stable and reliable when data deviate from idealized assumptions, such as the presence of outliers, heavier-tailed distributions, or violations of independence. These deviations can severely undermine classical procedures like the sample mean or least squares regression, which assume perfect normality and no contamination, leading to biased or inefficient estimates. The primary goal is to enhance the gross-error sensitivity of estimators, ensuring they perform well under realistic data conditions where contamination rates of 1-10% are common. The field traces its roots to early 19th-century critiques of least squares by astronomers like Bessel (1818) and Newcomb (1886), who noted empirical distributions often have heavier tails than the normal assumed by Gauss (1821). Modern robust statistics emerged in the mid-20th century, spurred by John Tukey's 1960 call for methods resilient to "bad" data points that could dominate analyses. Peter J. Huber laid foundational work with his 1964 paper on robust estimation of location parameters, introducing M-estimators that minimize a robust loss function to balance efficiency under normality with resistance to outliers. Frank Hampel further advanced the theory in 1968 by defining key robustness measures, including the influence function (quantifying an estimator's sensitivity to infinitesimal contamination at a point) and the breakdown point (the maximum fraction of corrupted data an estimator can tolerate before failing arbitrarily). The field expanded rapidly through the 1970s and 1980s, with Huber's 1981 book providing a comprehensive theoretical framework. Central to robust statistics are quantitative measures of performance: the point, where the achieves 50% (resisting up to half the data as ) compared to 0% for the , and the , which bounds the impact of any single observation to prevent undue . Notable methods include location estimators like the and trimmed (discarding extremes for resistance at the cost of some ), Huber's (using a piecewise linear loss for 95% asymptotic under ), and Tukey's biweight (a redescending estimator for stronger outlier rejection). For multivariate and settings, high- techniques such as the Least of Squares (LMS) and Minimum Determinant (MCD), developed by Peter Rousseeuw in the 1980s, achieve near-50% breakdown while detecting points. More advanced approaches like S-estimators and MM-estimators combine high breakdown with high , making them suitable for contaminated datasets. Robust methods are essential in fields like astronomy, , and , where data anomalies arise from measurement errors or model mismatches, improving reliability over classical alternatives. Ongoing developments address computational challenges in high dimensions and integrate robustness with for scalable applications.

Fundamentals

Introduction

Robust statistics encompasses statistical methods designed to yield reliable inferences even when the underlying data deviate from idealized model assumptions, such as the presence of outliers or incorrect distributional specifications. These techniques aim to provide stable and interpretable results by mitigating the impact of anomalous observations, which can otherwise lead to misleading conclusions in classical procedures like estimation. The field emerged in the 1960s as a response to the vulnerabilities of traditional statistics, pioneered by John W. Tukey and collaborators who emphasized the need for procedures resistant to real-world data imperfections. Tukey's advocacy for "resistant" or "robust" methods highlighted how classical estimators could fail dramatically under slight perturbations, prompting a shift toward exploratory and protective strategies. A core objective of robust statistics is to balance high —optimal performance under the assumed model—with to , ensuring estimators remain effective across a range of plausible data-generating processes. In the , Frank R. Hampel advanced this framework by developing tools to quantify and control the sensitivity of estimators to individual data points, laying foundational principles for modern robustness theory.

Definition and Prerequisites

Robust statistics refers to a branch of statistical theory and methodology that develops estimators and tests capable of maintaining desirable properties, such as efficiency and reliability, even when the underlying data distribution deviates moderately from the idealized model assumed in classical statistics. These deviations may include outliers, heavy-tailed distributions, or other forms of contamination. The performance of robust procedures is typically assessed through asymptotic properties in large samples, particularly by minimizing the worst-case asymptotic variance over a neighborhood of plausible distributions surrounding the target model. A key framework for evaluating robustness involves the contamination model, which formalizes potential deviations from the ideal . In this model, the observed distribution P_\epsilon is given by P_\epsilon = (1 - \epsilon) P + \epsilon Q, where P is the target (ideal) , Q is an arbitrary contaminating , and \epsilon \in [0, 1) represents the proportion of contamination, often taken to be small (e.g., \epsilon = 0.05 or $0.1) to reflect realistic scenarios. This model allows for the design of estimators that remain stable under limited gross errors while optimizing performance under P. In contrast to classical statistical methods, which assume that the data are drawn exactly from the specified model P and derive optimal procedures under that exact fit, robust methods explicitly account for possible contamination by Q, ensuring bounded degradation in performance. Classical approaches can exhibit extreme sensitivity to even a few outliers, leading to unreliable inferences, whereas robust alternatives prioritize stability across the \epsilon-neighborhood. To engage with robust statistics, foundational knowledge from probability and classical estimation theory is required. Basic concepts include the expectation of a random variable X, defined as E[X] = \int_{-\infty}^{\infty} x \, dF(x) for its cumulative distribution function F, and the variance \operatorname{Var}(X) = E[(X - E[X])^2], which measures dispersion around the mean. Classical estimators, such as the sample mean \bar{X} = n^{-1} \sum_{i=1}^n X_i for location and the sample standard deviation s = \sqrt{n^{-1} \sum_{i=1}^n (X_i - \bar{X})^2} for scale, assume independent and identically distributed (i.i.d.) observations from a known parametric family. Under these assumptions, the sample mean is unbiased (E[\bar{X}] = \mu) and consistent, meaning \bar{X} \xrightarrow{p} \mu as the sample size n \to \infty. Further prerequisites involve asymptotic central to both classical and robust . Consistency ensures that an \hat{\theta}_n converges in probability to the true parameter \theta as n \to \infty. Asymptotic , a consequence of the for i.i.d. data with finite variance, states that \sqrt{n} (\bar{X} - \mu) \xrightarrow{d} N(0, \sigma^2), providing a normal for in large samples. These form the basis for evaluating robust estimators, which extend them to contaminated settings while preserving similar guarantees under the target model P.

Measures of Robustness

Breakdown Point

The breakdown point quantifies the global robustness of a statistical by measuring the smallest fraction of that must be corrupted or replaced by arbitrary outliers to cause the to produce an unbounded or meaningless result, such as its value diverging to or its exceeding any predefined bound. This concept, originally developed for location estimators, serves as a key indicator of an estimator's reliability against gross errors in the . Two variants of the breakdown point are commonly distinguished: the finite-sample breakdown point, which evaluates robustness for a fixed sample size n by considering the minimal number of contaminated observations needed to spoil the estimator, and the asymptotic breakdown point, which assesses the limiting behavior as n \to \infty. The asymptotic version is particularly useful for theoretical comparisons, as it abstracts away from sample-specific details. The asymptotic breakdown point \eta^* of an estimator T for a distribution F is formally defined as \eta^* = \inf\left\{ \epsilon : \sup_Q \left| T\bigl( (1-\epsilon) F + \epsilon Q \bigr) - T(F) \right| = \infty \right\}, where \epsilon is the contamination proportion and the supremum is taken over all possible contaminating distributions Q. This represents the infimum of \epsilon values for which there exists some Q that drives the estimator arbitrarily far from its true value under the uncontaminated F. Illustrative examples highlight the range of breakdown points across estimators. The sample mean has a breakdown point of 0%, as even one can shift its value indefinitely. In contrast, the sample achieves the maximum possible breakdown point of 50%, requiring contamination of at least half the to cause breakdown. For the \alpha-trimmed mean, which discards the outer \alpha n / 2 observations from each end (total proportion \alpha trimmed) before computing the mean, the breakdown point is \alpha / 2 (or (\alpha / 2) \times 100\%). Despite its value as a global measure, the breakdown point has limitations: it focuses on worst-case catastrophic failure but overlooks finer-grained local robustness to small contaminations, and estimators designed for high breakdown points often exhibit reduced statistical efficiency when the data follow an ideal uncontaminated model like the normal distribution.

Influence Function and Sensitivity

The influence function serves as a key diagnostic tool in robust statistics, quantifying the local impact of an individual observation on an estimator viewed as a functional T of the underlying distribution F. It measures the infinitesimal change in T induced by contaminating F with a small proportion of mass at a specific point x. Formally, the influence function is defined as IF(x; T, F) = \lim_{\epsilon \to 0} \frac{T((1-\epsilon)F + \epsilon \delta_x) - T(F)}{\epsilon}, where \delta_x denotes the Dirac delta distribution concentrated at x. This expression captures the rate of change of T at F in the direction of \delta_x. The influence function's boundedness is central to assessing robustness: if |IF(x; T, F)| remains finite for all x, the estimator resists unbounded shifts from any single , unlike classical estimators such as the , whose influence function is unbounded and linear in x. The gross-error sensitivity \gamma^*(T, F) = \sup_x |IF(x; T, F)| provides a scalar summary of this maximum possible influence, with lower values indicating greater local robustness to point . For instance, the sample has a gross-error sensitivity of $1/(2f(\mu)), where f(\mu) is the at the median \mu, which is finite and typically small for symmetric distributions. The curve, obtained by plotting IF(x; T, F) against x, offers a visual representation of how varies across the of F. This curve highlights regions of high or low influence, aiding in the design of estimators that downweight extreme values; for example, a curve that rises initially but plateaus or declines for large |x| signals effective rejection. Desirable properties of the influence function include boundedness and a redescending shape for large deviations from the center, which promotes rejection of substantial residuals and enhances overall stability. Such redescending behavior aligns with the ψ-function in estimation procedures that diminish influence for outliers, contributing to qualitative robustness as defined by Hampel, where small perturbations in the data distribution lead to continuously small changes in the estimator. Qualitative robustness requires the estimator to be continuous with respect to weak convergence of distributions and the Prokhorov metric on probability measures. The arises as the Gâteaux derivative of T at F in the direction \delta_x - F, providing a first-order Taylor expansion for the contaminated functional: T((1-\epsilon)F + \epsilon \delta_x) \approx T(F) + \epsilon \cdot IF(x; T, F). This derivation, without requiring full differentiability assumptions, extends to von Mises functionals where T(F) = \int \phi(x) dF(x) for some integrable \phi, yielding IF(x; T, F) = \phi(x) - \int \phi(y) dF(y). While complementary to global measures like the breakdown point—which evaluates resistance to substantial contamination fractions—the influence function specifically probes sensitivity to infinitesimal point masses.

Core Methods

M-Estimators

M-estimators form a broad class of robust estimators introduced by Peter J. Huber in as a generalization of maximum likelihood estimators, designed to mitigate the influence of outliers in estimating parameters like location. For a sample X_1, \dots, X_n from a with location \theta and scale \sigma, the \hat{\theta} solves the estimating equation \sum_{i=1}^n \psi\left( \frac{X_i - \theta}{\sigma} \right) = 0, where \psi is an odd, non-decreasing function that bounds the influence of large residuals, unlike the unbounded \psi(u) = u in . Equivalently, \hat{\theta} minimizes the objective function \hat{\theta} = \arg\min_{\theta} \sum_{i=1}^n \rho\left( \frac{X_i - \theta}{\hat{\sigma}} \right), with \psi = \rho' and \hat{\sigma} a robust scale estimate, such as the . Under regularity conditions, including the existence of moments for \psi and \rho, M-estimators possess desirable asymptotic properties: they are consistent, asymptotically with rate \sqrt{n}, and their asymptotic variance is given by V = \frac{E[\psi^2(U)]}{[E[\psi'(U)]]^2}, where U follows the error . By tuning \psi, these estimators can achieve near-maximum at the —for instance, Huber's choice yields 95% efficiency while maintaining robustness—balancing bias and variance in contaminated models. Common \psi functions include Huber's, defined as \psi(u) = u for |u| \leq c and \psi(u) = c \cdot \operatorname{sign}(u) for |u| > c, which linearly weights small residuals but caps the influence of outliers at threshold c (often 1.345 for 95% efficiency). Another popular choice is Tukey's biweight, with \rho(u) = \frac{1}{6} \left(1 - (1 - u^2)^3 \right) for |u| \leq 1 (rescaled by c) and constant thereafter, leading to a redescending \psi(u) = u (1 - u^2)^2 for |u| < c that fully rejects extreme outliers by driving their weight to zero. This redescending behavior enhances robustness in heavy-tailed data but requires careful initialization to avoid local minima. Computing M-estimators typically involves iterative algorithms, with iteratively reweighted least squares (IRLS) being the standard approach: start with an initial \hat{\theta}^{(0)} (e.g., the median), compute weights w_i = \psi(r_i)/r_i where r_i = (X_i - \hat{\theta}^{(k)})/\hat{\sigma}, and update \hat{\theta}^{(k+1)} via weighted least squares until convergence. IRLS converges monotonically under bounded \rho and suitable starting values, often within 10-20 iterations for moderate samples, though it may require safeguards like trimming for redescending \psi.

Robust Parametric Approaches

Robust parametric approaches extend the M-estimation framework to structured parametric models, such as linear regression, where the goal is to estimate parameters under assumed functional forms while mitigating outlier effects. These methods generalize maximum likelihood estimation by replacing the log-likelihood with a robust loss function ρ that bounds the contribution of deviant observations, often applied to standardized residuals divided by a robust scale estimate σ. This adaptation ensures that the estimator remains consistent and asymptotically normal under mild contamination of the error distribution. In linear regression, the robust estimator \hat{\beta} solves an optimization problem that minimizes the sum of ρ over residuals: \hat{\beta} = \arg\min_{\beta} \sum_{i=1}^n \rho\left( \frac{y_i - x_i^T \beta}{\hat{\sigma}} \right), where \hat{\sigma} is typically a high-breakdown scale estimator like the median absolute deviation to avoid inflation by outliers. A classic implementation uses , where ρ(r) = r²/2 for |r| ≤ k and ρ(r) = k(|r| - k/2) otherwise, with the tuning constant k = 1.345 chosen to achieve 95% asymptotic efficiency relative to least squares under normality; this approach downweights but does not eliminate large residuals, providing a Gross Error Sensitivity of approximately 1.3σ. For enhanced robustness against leverage points and higher contamination levels, MM-estimators refine an initial high-breakdown S-estimator using a subsequent M-estimation step with a redescending ψ-function, such as Tukey's biweight, to achieve a breakdown point of up to 50% while retaining high efficiency (e.g., 85% or more at the normal). Developed by Yohai, this two-stage procedure ensures the final estimate is both affine-equivariant and computationally feasible for moderate dimensions, with the initial S-estimator minimizing a robust scale measure over subsets of the data. Key challenges in robust parametric estimation include sensitive initialization for non-convex losses, where poor starting values can lead to inconsistent solutions, and the selection of tuning constants for the ψ- and ρ-functions, which balance breakdown point against efficiency—for example, smaller constants enhance robustness but reduce efficiency under clean data. In , the serves as a robust initializer, but high-dimensional problems may require subsampling to manage computation, and constants like c=4.685 for ensure 95% efficiency while maintaining a 25% breakdown in the initial stage.

Applications and Extensions

Handling Outliers and Missing Values

Robust strategies for handling outliers begin with their detection, which can be achieved using robust residuals derived from preliminary robust fits, such as those obtained via , to identify observations that deviate significantly from the bulk of the data. Diagnostic plots, including scaled with robust measures of dispersion like the , further aid in visualizing departures from assumed distributions and flagging potential outliers. Once detected, outliers can be addressed through replacement techniques that mitigate their influence without complete removal. Winsorizing caps extreme values by replacing those below the α-quantile with the α-quantile value and those above the (1-α)-quantile with the (1-α)-quantile value, thereby preserving sample size while reducing outlier impact; a common choice is α=0.05, corresponding to the 5th and 95th percentiles. The winsorized mean is then computed as the arithmetic average of this adjusted dataset: \bar{x}_w = \frac{1}{n} \sum_{i=1}^n x_i^{(w)}, where x_i^{(w)} = q_{\alpha} if x_i < q_{\alpha}, x_i^{(w)} = x_i if q_{\alpha} \leq x_i \leq q_{1-\alpha}, and x_i^{(w)} = q_{1-\alpha} if x_i > q_{1-\alpha}, with q_{\alpha} denoting the α-quantile of the data. For imputation of missing or outlier-replaced values, robust central tendency measures like the median of complete cases provide a non-parametric alternative that resists contamination. Missing values pose a related challenge, often compounded by outliers in the observed data, and robust multiple imputation addresses this by generating multiple plausible imputations based on M-estimators fitted to complete cases, ensuring that the imputation model itself is insensitive to aberrant observations. This approach involves iteratively drawing from the posterior predictive distribution using robust location and scale estimates, then analyzing each imputed dataset separately before combining results via Rubin's rules. In scenarios with high outlier proportions, such as model fitting in noisy environments, the (RANSAC) algorithm offers a non-parametric method for robust by repeatedly sampling minimal subsets to hypothesize models, evaluating consensus via inlier counts, and selecting the model with the largest inlier set while discarding s. This iterative process, typically run for a fixed number of trials determined by desired confidence and outlier fraction, effectively isolates reliable data for in or geometric models.

Use in Machine Learning

In , robust statistics addresses key challenges posed by real-world data imperfections, such as noisy labels, adversarial attacks, and imbalanced distributions in large-scale datasets. Noisy labels, often arising from human annotation errors or automated labeling processes, can degrade model performance by misleading optimization toward incorrect patterns, particularly in deep neural networks where of amplifies errors. Adversarial attacks, which involve subtle perturbations to inputs that fool models while remaining imperceptible to humans, exploit vulnerabilities in gradient-based learning, leading to unreliable predictions in safety-critical applications like autonomous driving. Imbalanced data, prevalent in scenarios such as fraud detection or medical diagnostics, skews model training toward majority classes, resulting in poor for minority instances and heightened sensitivity to outliers. To mitigate these issues, robust techniques incorporate specialized loss functions and model modifications that downweight anomalous influences. Robust loss functions, such as the , blend quadratic behavior for small errors with linear penalties for large deviations, making them less sensitive to outliers than while remaining differentiable for in neural networks. These functions derive from M-estimators in robust statistics, adapting classical estimation principles to optimization. Outlier-robust support vector machines (SVMs) extend standard SVMs by integrating convex outlier ablation or algorithms during training, which identify and downweight non-support vectors affected by noise, thereby preserving margin-based separation in high-dimensional spaces. The Huber loss is defined as: L_\delta(y, f(x)) = \begin{cases} \frac{1}{2} (y - f(x))^2 & \text{if } |y - f(x)| < \delta \\ \delta |y - f(x)| - \frac{\delta^2}{2} & \text{otherwise} \end{cases} Practical examples illustrate these techniques' impact. decomposes data into low-rank and sparse components, enabling effective by isolating outliers as the sparse residual, which has proven valuable in network traffic monitoring where it achieves low false-positive rates on packet data. In , certified robustness provides provable guarantees against adversarial perturbations, with post-2010s advances like randomized smoothing and interval bound propagation scaling to ImageNet-sized models and diverse architectures, ensuring epsilon-bounded resilience in classifiers. Recent developments through 2025 have integrated robust statistics with federated learning to enhance privacy-preserving robustness across distributed systems. In federated settings, where models aggregate updates from decentralized clients without sharing raw data, robust aggregation rules—such as trimmed means or median-based methods—counter poisoning attacks and non-IID data heterogeneity, improving convergence and accuracy in meta-learning tasks like personalized recommendation. Systematic reviews highlight that these integrations, often combining contrastive learning with outlier detection, mitigate label noise and Byzantine faults while complying with differential privacy constraints, leading to improved performance on heterogeneous benchmarks.

Comparisons to Classical Statistics

Classical estimators, such as the maximum likelihood (MLE), achieve optimal asymptotic when the underlying model assumptions are satisfied. For instance, under a , the sample serves as the MLE for the and attains an asymptotic relative (ARE) of 1.0 relative to itself. In contrast, robust estimators trade off some for resistance to deviations from the model. The sample median, a canonical robust location , exhibits an ARE of approximately 0.64 compared to the under normality, though its improves significantly under heavier-tailed distributions, reaching about 0.96 for a with 5 . Classical approaches rely on strict adherence to parametric assumptions, such as exact of errors for the validity of the or . Violations, even minor ones like outliers, can lead to substantial or inefficiency. Robust methods, however, are formulated to tolerate such deviations through models like ε-contamination, where the observed is (1-ε) times the ideal model plus ε times an arbitrary contaminating , allowing reliable for small ε (typically up to 0.1–0.2). A key in robust statistics is between robustness and computational cost; highly robust procedures often require iterative optimization, increasing runtime compared to closed-form classical solutions like the . For example, the Huber M-estimator solves a problem via numerical methods, potentially demanding more resources than ordinary . Additionally, achieving high robustness, such as a breakdown point near 0.5, typically reduces under ideal conditions, whereas estimators like those based on the t-distribution strike a moderate balance, offering reasonable robustness against moderate while maintaining higher than the for symmetric heavy-tailed errors. Classical methods are appropriate for controlled environments with clean data satisfying model assumptions, ensuring maximal precision. Robust methods should be preferred in real-world scenarios where data messiness is common; literature on and statistics notes that most real-world datasets contain outliers, which can severely classical estimates if unaddressed. The following table summarizes key comparisons for location estimators under the normal distribution, highlighting the trade-offs in breakdown point (maximum fraction of outliers tolerable before arbitrary breakdown) and ARE relative to the :
EstimatorBreakdown PointARE (normal)
Mean01.0
0.50.64
Huber M-estimator0.5≈0.95
These values illustrate how the excels in efficiency but fails with any outliers, the provides maximal robustness at the cost of efficiency, and the Huber estimator offers a practical with high efficiency and moderate robustness.

Historical Development

The origins of robust statistics can be traced to the late , when astronomers and statisticians began addressing the sensitivity of classical estimators to outliers in observational data. Simon Newcomb's 1886 discussion provided one of the first modern frameworks for robust estimation, introducing the use of mixture models of normal densities to account for contaminated data distributions. Earlier ideas, such as the as a resistant measure of , had roots in 18th-century , with precursors in the works of Laplace and others who recognized the limitations of the under non-normal errors, though Carl Friedrich Gauss's 1809 development of emphasized assumptions of equal accuracy that later highlighted the need for robustness. The 20th century saw a renewed push toward robust methods, particularly through John Tukey's exploratory data analysis in the 1960s, which emphasized graphical techniques and resistant summaries to uncover structure in potentially messy data. Tukey's seminal 1960 paper, "A Survey of Sampling from Contaminated Distributions," formalized the concept of contamination models and inspired systematic investigations into robustness, marking a shift from assuming perfect data to designing procedures resilient to deviations. Key figures shaped the theoretical foundations in the mid-20th century. Peter J. Huber introduced M-estimators in 1964 as a generalization of maximum likelihood for handling gross errors, providing a framework for location estimation under contamination. Frank R. Hampel advanced this in 1974 with the , a tool to quantify the impact of individual observations on estimators and guide the design of qualitatively robust procedures. In the 1980s, Peter J. Rousseeuw developed high-breakdown-point methods, such as least median of squares regression in 1984 and minimum covariance determinant in 1985, which could withstand up to nearly 50% contamination without total failure. Milestones in the 1970s included dedicated academic gatherings, such as the 1970 Princeton seminar on robust statistics, which brought together leading researchers to discuss stability under model misspecification. The decade also saw the emergence of robustness as a core theme in statistical conferences, culminating in events like the 1979 symposium that disseminated early results. By the 1990s, practical implementation advanced with software integration; introduced robust libraries featuring M-estimators and regression tools, enabling widespread application in statistical computing environments. In the from the to 2025, robust statistics has integrated with challenges, developing scalable estimators for high-dimensional settings, such as median-of-means for sub-Gaussian robustness without strong moment assumptions. Bayesian robustness gained traction, with partial prior specifications and sensitivity analyses addressing model uncertainty in complex posteriors, as explored in foundational overviews from the mid-2000s. Post-2020, the field extended to , with seminal works on adversarial robustness using certified defenses and to mitigate vulnerabilities in neural networks, exemplified by efficient training methods achieving state-of-the-art protection against white-box attacks. Computational advances post-2010, including polynomial-time algorithms for high-breakdown covariance estimation, have been supported by open-source tools; in , libraries like statsmodels.robust (introduced around 2012) and the more recent RobPy package provide accessible implementations for and multivariate analysis.

References

  1. [1]
    [PDF] Robust statistics: A brief introduction and overview
    Mar 12, 2001 · Robust statistics is the stability theory of statistical procedures. It systematically investigates the effects of deviations from modelling ...Missing: definition | Show results with:definition
  2. [2]
    [PDF] Robust Statistics
    Huber, P. J. (1981) Robust Statistics. New York: John Wiley and Sons. Iglewicz, B. (1983) Robust scale estimators and confidence intervals for location. In ...<|control11|><|separator|>
  3. [3]
    Robust Estimation of a Location Parameter - Semantic Scholar
    Robust Estimation of a Location Parameter · P. J. Huber · Published 1 March 1964 · Mathematics · Annals of Mathematical Statistics.Missing: seminal | Show results with:seminal<|separator|>
  4. [4]
    Robust Statistics, 2nd Edition - Wiley
    The first edition was written by Peter J. Huber who is viewed as one of the leading developers of robust statistics and as on of the sharpest innovators in the ...
  5. [5]
  6. [6]
    A survey of robust statistics | Statistical Methods & Applications
    Jan 3, 2007 · We argue that robust statistics has multiple goals, which are not always aligned. Robust thinking grew out of data analysis.
  7. [7]
    [PDF] Robust Statistics
    Statistics is the science of extracting useful information from data, and a statistical model is used to provide a useful approximation to some of the.
  8. [8]
    [PDF] Introduction to Robust Statistics Elvezio Ronchetti Department of ...
    • primary goal is the development of pro- cedures which are still reliable and rea- sonably efficient under small deviations from the model, i.e. when the ...Missing: key | Show results with:key
  9. [9]
    John Tukey and Robustness - jstor
    This article discusses some of these contributions, with a special emphasis on those that led to the development of robust methods and data exploration. In view ...Missing: origins | Show results with:origins
  10. [10]
    [PDF] The Changing History of Robustness
    Jan 1, 2012 · This essay, a reflection upon the changing views of robust statistics from the euphoria of the 1960s to the present day, was given as a keynote ...Missing: definition | Show results with:definition
  11. [11]
    The Influence Curve and Its Role in Robust Estimation - jstor
    [10] Hampel, Frank R., "Contributions to the Theory of Robust. Estimation," Unpublished Ph.D. thesis, University of Cali- fornia, Berkeley, September 1968.
  12. [12]
    Robust Estimation of a Location Parameter - Project Euclid
    March, 1964 Robust Estimation of a Location Parameter. Peter J. Huber · DOWNLOAD PDF + SAVE TO MY LIBRARY. Ann. Math. Statist. 35(1): 73-101 (March, 1964). DOI ...
  13. [13]
    Outliers in Time Series - Fox - 1972 - Royal Statistical Society - Wiley
    Summary. Two models are considered for outliers and their effects in time series. Likelihood ratio and approximate likelihood ratio criteria are derived for ...
  14. [14]
    [PDF] ANOVA – The Effect of Outliers - DiVA portal
    One sizeable outlier may cause the sample mean to deviate from the true mean, and causes a substantial variation. Thus, the variance of the sample mean must not ...<|control11|><|separator|>
  15. [15]
    [PDF] Robust Statistical Methods for Automated Outlier Detection
    In an effort to balance the total squared error, the sample mean is forced to overcompensate for the one outlier. Consider now changing the form of the error ...
  16. [16]
    Editorial, special issue on “Advances in Robust Statistics” - PMC
    Anomaly detection techniques based on robust statistics are also applied by Baesens et al. [2] to detect fraud in a real payment transactions data set from a ...
  17. [17]
    A Review of Outlier Detection and Robust Estimation Methods for ...
    Feb 5, 2023 · As high dimensional data sets are expected to include some outliers, robust estimation methods are required for automatic analysis on these ...
  18. [18]
    Chapter 12 Robust summaries | Introduction to Data Science - rafalab
    The median, defined as the value for which half the values are smaller and the other half are bigger, is robust to such outliers. No matter how large we make ...
  19. [19]
    1.3.5.6. Measures of Scale - Information Technology Laboratory
    The median absolute deviation and the interquartile range are estimates of scale that have robustness of validity. However, they are not particularly strong ...
  20. [20]
    [PDF] THE BREAKDOWN POINT — EXAMPLES AND ...
    The breakdown point is one of the most popular measures of robustness of a statistical procedure. Originally introduced for location functionals (Hampel, 1968, ...
  21. [21]
    Comprehensive Definitions of Breakdown Points for Independent ...
    For example, Hampel's (1971) original definition uses the absolute bias as the criterion function and ∞ (or the edge of the parameter space) as the critical ...Introduction · Definition of breakdown · Examples: location · Examples: time series
  22. [22]
    (PDF) The Breakdown Point—Examples and Counterexamples
    Aug 7, 2025 · ... The robustness of an estimator against outliers is commonly assessed through the breakdown point [61] . This index is defined as the largest ...Missing: seminal | Show results with:seminal
  23. [23]
    Robust Statistics: The Approach Based on Influence Functions
    Robust Statistics: The Approach Based on Influence Functions. Author(s): Frank R. Hampel, Elvezio M. Ronchetti, Peter J. Rousseeuw, Werner A. Stahel.
  24. [24]
    A General Qualitative Definition of Robustness - jstor
    1971, Vol. 42, No. 6, 1887-1896. A GENERAL QUALITATIVE DEFINITION OF ROBUSTNESS'. BY FRANK R. HAMPEL. University of California, Berkeley, and University of ...
  25. [25]
    The Fitting of Power Series, Meaning Polynomials, Illustrated on ...
    Apr 9, 2012 · Technometrics Volume 16, 1974 - Issue 2 · Submit an article Journal ... The Fitting of Power Series, Meaning Polynomials, Illustrated on Band- ...Missing: biweight | Show results with:biweight
  26. [26]
    Robust regression using iteratively reweighted least-squares
    Jun 27, 2007 · We will review a number of different computational approaches for robust linear regression but focus on one—iteratively reweighted least-squares ...
  27. [27]
    Robust Regression: Asymptotics, Conjectures and Monte Carlo
    Maximum likelihood type robust estimates of regression are defined and their asymptotic properties are investigated both theoretically and empirically.
  28. [28]
    Robust Statistics - Wiley Online Library
    Robust Statistics ; First published:29 January 2009 ; Print ISBN:9780470129906 | ; Online ISBN:9780470434697 | ; DOI:10.1002/9780470434697 ; Book ...
  29. [29]
    High Breakdown-Point and High Efficiency Robust Estimates for ...
    The MM-estimates are defined by a three-stage procedure. In the first stage an initial regression estimate is computed which is consistent robust and with high ...
  30. [30]
    Do Robust Estimators Work with Real Data? - Project Euclid
    This paper presents a comparison of the performances of eleven estimators using real data sets. Twenty sets of data from 1761 determinations of the parallax of ...
  31. [31]
    Outliers and robust procedures in some chemometric applications
    The control-chart-type quantile-quantile (Q-Q) plot of Mahalanobis distances provides a formal graphical multiple outlier identification procedure. Also the ...
  32. [32]
    Random sample consensus: a paradigm for model fitting with ...
    Jun 1, 1981 · A major portion of this paper describes the application of RANSAC to the Location Determination Problem (LDP): Given an image depicting a ...
  33. [33]
    [PDF] Effective and Robust Adversarial Training against Data and Label ...
    May 7, 2024 · ERAT uses hybrid adversarial training with multiple perturbations and semi-supervised learning to handle data and label corruptions, and remove ...
  34. [34]
    Robustness in deep learning models for medical diagnostics
    Nov 8, 2024 · Robust training involves incorporating noise or perturbations into the training dataset, making the model more resistant to adversarial attacks ...
  35. [35]
    Outlier Path: A Homotopy Algorithm for Robust SVM
    Although improving the robustness of SVM has been investigated for long time, robust SVM (RSVM) learning still poses two major challenges: obtaining a good ( ...
  36. [36]
    Robust PCA for Anomaly Detection in Cyber Networks - arXiv
    Jan 4, 2018 · This paper uses Robust PCA (RPCA) to detect cyber-network anomalies using network packet data, achieving low false-positive rates.
  37. [37]
    [2009.04131] SoK: Certified Robustness for Deep Neural Networks
    This paper systematizes certifiably robust approaches for deep neural networks, addressing their vulnerability to adversarial attacks, and provides a benchmark.
  38. [38]
    Full article: Robust Inference for Federated Meta-Learning
    In this article, we consider a general robust inference framework for federated meta-learning of data from multiple sites, enabling statistical inference for ...
  39. [39]
    A Systematic Literature Review of Robust Federated Learning
    Particularly, robust FL seeks to mitigate the risks posed by malicious clients, noisy updates, and inconsistent data by employing advanced aggregation methods, ...
  40. [40]
    [PDF] A Short Course on Robust Statistics
    • GES = ∞, i.e. unbounded influence. • Outlier 1: highly influential for any ψ function. • Outlier 2: highly influential for monotonic but not redescending ψ.Missing: key goals
  41. [41]
    [PDF] Simon Newcomb, Percy Daniell, and the History of Robust ... - DTIC
    May 30, 2023 · Simon tlewcomb (1886) provided the first sound, modern approach to robust estimation, including the first use of mixtures of normal densities as.
  42. [42]
    [PDF] Robust Statistics: A Brief Introduction and Overview
    Mar 12, 2001 · Robust statistics is the stability theory of statistical procedures. It systematically investigates the effects of deviations from modelling ...
  43. [43]
    of Robust Estimation 1885-1920 - jstor
    Yet most of the early work in mathematical statistics was obsessed with "proving" the method of least squares, either starting with the assumption that the ...Missing: roots | Show results with:roots
  44. [44]
    [PDF] S-PLUS 6 Robust Library
    This release contains the following contributed S-PLUS code: robust spline functions from Elvezio Ronchetti and Eva Cantoni; robust Cp functions from Robert ...
  45. [45]
    A Theoretical Review of Modern Robust Statistics
    Mar 7, 2025 · Robust statistics is a fairly mature field that dates back to the early 1960s, with many foundational concepts having been developed in the ...
  46. [46]
    Robust Bayesian Analysis - ScienceDirect.com
    We provide an overview of robust Bayesian analysis with emphasis on foundational, decision oriented and computational approaches. Common types of robustness ...
  47. [47]
    [PDF] Robust Deep Reinforcement Learning against Adversarial ...
    We significantly improve the robustness of PPO, DDPG and DQN agents under a suite of strong white box adversarial attacks, including new attacks of our own.Missing: seminal post-