Robust Estimation of a Location Parameter

1964/03/01 by Peter J. Huber · 277 citations
Mathematics · Decision Sciences · #Advanced Statistical Methods and Models #Advanced Statistical Process Monitoring #Statistical Methods and Inference

paper · pdf · doi:10.1214/aoms/1177703732

Abstract

This paper contains a new approach toward a theory of robust estimation; it treats in detail the asymptotic theory of estimating a location parameter for contaminated normal distributions, and exhibits estimators--intermediaries between sample mean and sample median--that are asymptotically most robust (in a sense to be specified) among all translation invariant estimators. For the general background, see Tukey (1960) (p. 448 ff.) Let x₁, ⋯, x_n be independent random variables with common distribution function F(t - ξ). The problem is to estimate the location parameter ξ, but with the complication that the prototype distribution F(t) is only approximately known. I shall primarily be concerned with the model of indeterminacy F = (1 - ε)Φ + ε H, where 0 \leqq ε < 1 is a known number, Φ(t) = (2π)^-(1)/(2) ∫^t_-∞ exp(-(1)/(2)s²) ds is the standard normal cumulative and H is an unknown contaminating distribution. This model arises for instance if the observations are assumed to be normal with variance 1, but a fraction ε of them is affected by gross errors. Later on, I shall also consider other models of indeterminacy, e.g., sup_t |F(t) - Φ(t)| \leqq ε. Some inconvenience is caused by the fact that location and scale parameters are not uniquely determined: in general, for fixed ε, there will be several values of ξ and σ such that sup_t|F(t) - Φ((t - ξ)/σ)| \leqq ε, and similarly for the contaminated case. Although this inherent and unavoidable indeterminacy is small if ε is small and is rather irrelevant for practical purposes, it poses awkward problems for the theory, especially for optimality questions. To remove this difficulty, one may either (i) restrict attention to symmetric distributions, and estimate the location of the center of symmetry (this works for ξ but not for σ); or (ii) one may define the parameter to be estimated in terms of the estimator itself, namely by its asymptotic value for sample size n → ∞; or (iii) one may define the parameters by arbitrarily chosen functionals of the distribution (e.g., by the expectation, or the median of F). All three possibilities have unsatisfactory aspects, and I shall usually choose the variant which is mathematically most convenient. It is interesting to look back to the very origin of the theory of estimation, namely to Gauss and his theory of least squares. Gauss was fully aware that his main reason for assuming an underlying normal distribution and a quadratic loss function was mathematical, i.e., computational, convenience. In later times, this was often forgotten, partly because of the central limit theorem. However, if one wants to be honest, the central limit theorem can at most explain why many distributions occurring in practice are approximately normal. The stress is on the word "approximately." This raises a question which could have been asked already by Gauss, but which was, as far as I know, only raised a few years ago (notably by Tukey): What happens if the true distribution deviates slightly from the assumed normal one? As is now well known, the sample mean then may have a catastrophically bad performance: seemingly quite mild deviations may already explode its variance. Tukey and others proposed several more robust substitutes--trimmed means, Winsorized means, etc.--and explored their performance for a few typical violations of normality. A general theory of robust estimation is still lacking; it is hoped that the present paper will furnish the first few steps toward such a theory. At the core of the method of least squares lies the idea to minimize the sum of the squared "errors," that is, to adjust the unknown parameters such that the sum of the squares of the differences between observed and computed values is minimized. In the simplest case, with which we are concerned here, namely the estimation of a location parameter, one has to minimize the expression ∑_i (x_i - T)²; this is of course achieved by the sample mean T = ∑_i x_i/n. I should like to emphasize that no loss function is involved here; I am only describing how the least squares estimator is defined, and neither the underlying family of distributions nor the true value of the parameter to be estimated enters so far. It is quite natural to ask whether one can obtain more robustness by minimizing another function of the errors than the sum of their squares. We shall therefore concentrate our attention to estimators that can be defined by a minimum principle of the form (for a location parameter): T = T_n(x₁, ⋯, x_n) minimizes ∑_i ρ(x_i - T), where ρ is a non-constant function. Of course, this definition generalizes at once to more general least squares type problems, where several parameters have to be determined. This class of estimators contains in particular (i) the sample mean (ρ(t) = t²), (ii) the sample median (ρ(t) = |t|), and more generally, (iii) all maximum likelihood estimators (ρ(t) = -log f(t), where f is the assumed density of the untranslated distribution). These (M)-estimators, as I shall call them for short, have rather pleasant asymptotic properties; sufficient conditions for asymptotic normality and an explicit expression for their asymptotic variance will be given. How should one judge the robustness of an estimator T_n(x) = T_n(x₁, ⋯, x_n)? Since ill effects from contamination are mainly felt for large sample sizes, it seems that one should primarily optimize large sample robustness properties. Therefore, a convenient measure of robustness for asymptotically normal estimators seems to be the supremum of the asymptotic variance (n → ∞) when F ranges over some suitable set of underlying distributions, in particular over the set of all F = (1 - ε)Φ + ε H for fixed ε and symmetric H. On second thought, it turns out that the asymptotic variance is not only easier to handle, but that even for moderate values of n it is a better measure of performance than the actual variance, because (i) the actual variance of an estimator depends very much on the behavior of the tails of H, and the supremum of the actual variance is infinite for any estimator whose value is always contained in the convex hull of the observations. (ii) If an estimator is asymptotically normal, then the important central part of its distribution and confidence intervals for moderate confidence levels can better be approximated in terms of the asymptotic variance than in terms of the actual variance. If we adopt this measure of robustness, and if we restrict attention to (M)-estimators, then it will be shown that the most robust estimator is uniquely determined and corresponds to the following ρ:ρ(t) = (1)/(2)t² for |t| < k, ρ(t) = k|t| - (1)/(2)k² for |t| \geqq k, with k depending on ε. This estimator is most robust even among all translation invariant estimators. Sample mean (k = ∞) and sample median (k = 0) are limiting cases corresponding to ε = 0 and ε = 1, respectively, and the estimator is closely related and asymptotically equivalent to Winsorizing. I recall the definition of Winsorizing: assume that the observations have been ordered, x₁ \leqq x₂ \leqq ⋯ \leqq x_n, then the statistic T = n^-1(gx_{g + 1} + x_{g + 1} + x_{g + 2} + ⋯ + x_{n - h} + hx_{n - h}) is called the Winsorized mean, obtained by Winsorizing the g leftmost and the h rightmost observations. The above most robust (M)-estimators can be described by the same formula, except that in the first and in the last summand, the factors x_{g + 1} and x_{n - h} have to be replaced by some numbers u, v satisfying x_g \leqq u \leqq x_{g + 1} and x_{n - h} \leqq v \leqq x_{n - h + 1}, respectively; g, h, u and v depend on the sample. In fact, this (M)-estimator is the maximum likelihood estimator corresponding to a unique least favorable distribution F₀ with density f₀(t) = (1 - ε)(2π)^-(1)/(2)e^-ρ(t). This f₀ behaves like a normal density for small t, like an exponential density for large t. At least for me, this was rather surprising--I would have expected an f₀ with much heavier tails. This result is a particular case of a more general one that can be stated roughly as follows: Assume that F belongs to some convex set C of distribution functions. Then the most robust (M)-estimator for the set C coincides with the maximum likelihood estimator for the unique F₀ ε C which has the smallest Fisher information number I(F) = ∫ (f'/f)²f dt among all F ε C. Miscellaneous related problems will also be treated: the case of non-symmetric contaminating distributions; the most robust estimator for the model of indeterminacy sup_t|F(t) - Φ(t)| \leqq ε; robust estimation of a scale parameter; how to estimate location, if scale and ε are unknown; numerical computation of the estimators; more general estimators, e.g., minimizing ∑_{i < j} ρ(x_i - T, x_j - T), where ρ is a function of two arguments. Questions of small sample size theory will not be touched in this paper.

Cited by

• On Some Robust Estimates of Location
• Student of Games: A unified learning algorithm for both perfect and imperfect information games
• DeepStack: Expert-level artificial intelligence in heads-up no-limit poker
• DiffractioNet: Deep-learning seismic and GPR diffraction separation
• How Censorship in China Allows Government Criticism but Silences Collective Expression
• Geostatistical Tools for Modeling and Interpreting Ecological Spatial Dependence
• ECG-based convolutional neural network in pediatric obstructive sleep apnea diagnosis
• Performance of the stochastic MV-PURE estimator in highly noisy settings
• Electric transmission value and its drivers in United States power markets
• Democracy Does Cause Growth
• Robust tools for the imperfect world
• Investor Protection, Equity Returns, and Financial Globalization
• An opponent striatal circuit for distributional reinforcement learning
• A distributionally robust perspective on uncertainty quantification and chance constrained programming
• Proton range verification through prompt gamma-ray spectroscopy
• Returns to education and wages in Turkey: robust and resistant regression
• Computational experiments with cellular-automata generated images reveal intrinsic limitations of convolutional neural networks on pattern recognition tasks
• Statistical inference in two-sample summary-data Mendelian randomization using robust adjusted profile score
• Dual divergence estimators and tests: Robustness results
• Gaussian and robust Kronecker product covariance estimation: Existence and uniqueness
• Robust machine learning by median-of-means: Theory and practice
• Large covariance estimation through elliptical factor models
• Confidence regions and minimax rates in outlier-robust estimation on the\n probability simplex
• Climate-induced interannual variability in length–weight ratios of the pink abalone ( Haliotis corrugata ) as an indicator for fisheries management
• Robust Dynamic Assortment Optimization in the Presence of Outlier Customers
• Valid Post-selection Inference in Assumption-lean Linear Regression
• RaD-VIO: Rangefinder-aided Downward Visual-Inertial Odometry
• Robust Mean Estimation in High Dimensions via ℓ₀ Minimization
• Robust Regression via Mutivariate Regression Depth
• Learning Deep Neural Networks under Agnostic Corrupted Supervision
• Baseline Estimation of Commercial Building HVAC Fan Power Using Tensor Completion
• Boosting Functional Regression Models with FDboost
• Outlier-Robust Sparse Estimation via Non-Convex Optimization
• Robust penalized empirical likelihood in high dimensional longitudinal data analysis
• Roughly Collected Dataset for Contact Force Sensing Catheter
• Tailored Bayes: a risk modelling framework under unequal misclassification costs
• Differential privacy and robust statistics in high dimensions
• On the Equivalence of Inexact Proximal ALM and ADMM for a Class of Convex Composite Programming
• PL-VINS: Real-Time Monocular Visual-Inertial SLAM with Point and Line Features
• An ℓ_∞ Eigenvector Perturbation Bound and Its Application to Robust Covariance Estimation
• Robust Training in High Dimensions via Block Coordinate Geometric Median Descent
• GRB Redshift Classifier to Follow-up High-Redshift GRBs Using Supervised Machine Learning
• Digital Sleep-Wake Cycle Metrics and Dementia Prediction in Older Adults
• Optimal Mean Estimation without a Variance
• Equivariant Q Learning in Spatial Action Spaces
• Me, My Self (Concept), and I – Disentangling the Link Between Narcissism and the Impostor Self-Concept
• Estimation bounds and sharp oracle inequalities of regularized\n procedures with Lipschitz loss functions
• From Softmax to Sparsemax: A Sparse Model of Attention and Multi-Label Classification
• Vehicular Network Slicing for Reliable Access and Deadline-Constrained Data Offloading: A Multi-Agent On-Device Learning Approach
• Simultaneous penalized M-estimation of covariance matrices using\n geodesically convex optimization
• Reducibility and Statistical-Computational Gaps from Secret Leakage
• Deep Huber quantile regression networks
• TomoCPT : a generalizable model for 3D particle detection and localization in cryo-electron tomograms
• SimNet: Enabling Robust Unknown Object Manipulation from Pure Synthetic Data via Stereo
• Being Robust (in High Dimensions) Can Be Practical
• The determinants of corporate board size and composition: An empirical analysis
• On Empirical Risk Minimization with Dependent and Heavy-Tailed Data
• Learning from MOM's principles: Le Cam's approach
• Stochastic Smoothing for Nonsmooth Minimizations: Accelerating SGD by Exploiting Structure
• Location estimation for symmetric log-concave densities
• Robust optima and tolerance ranges of biological indicators: a new method to identify sentinels of global warming
• Fast Differentiable Sorting and Ranking
• Capsule Networks for Protein Structure Classification and Prediction
• Mask2CAD: 3D Shape Prediction by Learning to Segment and Retrieve
• Privilege or Pathway? Latent Profiles and Unequal Academic Returns from High-Impact Practices in Higher Education
• Defending Regression Learners Against Poisoning Attacks
• A General Optimization-based Framework for Local Odometry Estimation with Multiple Sensors
• Reliable and Efficient Image Cropping: A Grid Anchor based Approach
• Out-of-sample error estimate for robust M-estimators with convex penalty
• LqRT: Robust Hypothesis Testing of Location Parameters using Lq-Likelihood-Ratio-Type Test in Python
• Studying Very Low Resolution Recognition Using Deep Networks
• Variable transformations in consistent loss functions
• Unsupervised Learning of Solutions to Differential Equations with Generative Adversarial Networks
• Divergence-Free Shape Interpolation and Correspondence
• Bandits with heavy tail
• Learning spatial hearing via innate mechanisms
• Firms’ Inflation Expectations in a Monetary Union
• Pointwise Adaptive M-estimation in Nonparametric Regression
• Outlier-robust sparse/low-rank least-squares regression and robust\n matrix completion
• Robust and flexible inference for the covariate-specific ROC curve
• Inference with penalized likelihood
• High Dimensional Robust Sparse Regression
• Integrating rapid assessment, variable probability sampling, and machine learning to improve accuracy and consistency in mapping local spatial distribution of plant species richness
• Outlier-robust estimation of a sparse linear model using ℓ₁-penalized Huber's M-estimator
• Block-wise Minimization-Majorization algorithm for Huber's criterion: sparse learning and applications
• Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation
• DebiNet: Debiasing Linear Models with Nonlinear Overparameterized Neural\n Networks
• Average-Case Lower Bounds for Learning Sparse Mixtures, Robust\n Estimation and Semirandom Adversaries
• Making and Evaluating Point Forecasts
• Deep geometric knowledge distillation with graphs
• Video2GIF: Automatic Generation of Animated GIFs from Video
• Optimal Design of the Adaptive Normalized Matched Filter Detector
• Benchmarking Batch Deep Reinforcement Learning Algorithms
• KKT Conditions, First-Order and Second-Order Optimization, and Distributed Optimization: Tutorial and Survey
• Richard Hooker on the eucharist : A commentary on the Laws V.67
• Quantifying Error in the Presence of Confounders for Causal Inference
• Robust Aggregation for Federated Learning
• Robust estimation of location and concentration parameters for the von Mises-Fisher distribution
• Empirical Strategy for Stretching Probability Distribution in Neural-network-based Regression
• A simple recipe for making accurate parametric inference in finite sample
• Robust Learning of Fixed-Structure Bayesian Networks
• Advances in Inference and Representation for Simultaneous Localization\n and Mapping
• Robust Nonparametric Regression with Deep Neural Networks
• Better Algorithms for Stochastic Bandits with Adversarial Corruptions
• Robust mixture modelling using the t distribution
• Uncertainty Quantification for Online Learning and Stochastic Approximation via Hierarchical Incremental Gradient Descent
• Simple and practical algorithms for ℓ_p-norm low-rank approximation
• Neural mechanisms underlying the hierarchical construction of perceived aesthetic value
• Robust Estimation under Heavy Contamination using Enlarged Models
• High finite-sample efficiency and robustness based on\n distance-constrained maximum likelihood
• High Dimensional Robust M-Estimation: Asymptotic Variance via\n Approximate Message Passing
• Enveloped Huber Regression
• Distributed Cooperative Localization in Wireless Sensor Networks without NLOS Identification
• A deep architecture for unified aesthetic prediction
• Instance Segmentation GNNs for One-Shot Conformal Tracking at the LHC
• Practical selection of SVM parameters and noise estimation for SVM regression
• Measuring Average Treatment Effect from Heavy-tailed Data
• A Reinforcement Learning Based Encoder-Decoder Framework for Learning Stock Trading Rules
• VN-EGNN: E(3)- and SE(3)-Equivariant Graph Neural Networks with Virtual Nodes Enhance Protein Binding Site Identification
• Revisiting Rainbow: Promoting more Insightful and Inclusive Deep\n Reinforcement Learning Research
• Robust Federated Learning by Mixture of Experts
• Optimal Robust Learning of Discrete Distributions from Batches
• Duality in RKHSs with Infinite Dimensional Outputs: Application to Robust Losses
• Convergence of online mirror descent
• A Learning-from-noise Dilated Wide Activation Network for denoising Arterial Spin Labeling (ASL) Perfusion Images
• Beyond Parity: Fairness Objectives for Collaborative Filtering
• A flexible trajectory estimation methodology for kinematic laser scanning
• A survey of cross-validation procedures for model selection
• Structure from Motion on a Sphere
• Implicit Quantile Networks for Distributional Reinforcement Learning
• LEAST TRIMMED SQUARES: NUISANCE PARAMETER FREE ASYMPTOTICS
• 3D Objectness Estimation via Bottom-up Regret Grouping
• Multi-Label Segmentation via Residual-Driven Adaptive Regularization
• Campolina: A Deep Neural Framework for Accurate Segmentation of Nanopore Signals
• Systematic analysis of drug combinations against Gram-positive bacteria
• A nonasymptotic law of iterated logarithm for general M-estimators
• CRIMED: Lower and Upper Bounds on Regret for Bandits with Unbounded Stochastic Corruption
• NEBULA is a fast negative binomial mixed model for differential or co-expression analysis of large-scale multi-subject single-cell data
• Statistical Analysis of Nearest Neighbor Methods for Anomaly Detection
• Online Robust Regression via SGD on the l1 loss
• Statistical Query Lower Bounds for List-Decodable Linear Regression
• A General Decision Theory for Huber's ε-Contamination Model
• Learning theory of regularized Huber regression
• Mixture Models, Robustness, and Sum of Squares Proofs
• Compressed matched filter for non-Gaussian noise
• Dynamic Neural Turing Machine with Soft and Hard Addressing Schemes
• Robust tightly coupled pose estimation based on monocular vision, inertia, and wheel speed
• Robustness and Tractability for Non-convex M-estimators
• Sample-based Distributional Policy Gradient
• On the Runtime-Efficacy Trade-off of Anomaly Detection Techniques for Real-Time Streaming Data
• Perspective Functions: Properties, Constructions, and Examples
• Empirical Asset Pricing via Machine Learning
• Robustness through Data Augmentation Loss Consistency
• A visual introduction to Gaussian Belief Propagation
• Neighborhoods as Nuisance Parameters? Robustness vs. Semiparametrics
• On Tilted Losses in Machine Learning: Theory and Applications
• Robust and Differentially Private Mean Estimation
• Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud
• Pose Estimation for Omni-directional Cameras using Sinusoid Fitting
• Cost-Oriented Load Forecasting
• Improved Natural Language Generation via Loss Truncation
• Privacy-Aware Recommender Systems Challenge on Twitter's Home Timeline
• Generalization of Eight Methods for Determining R in the Ideal Gas Law
• OSQP: an operator splitting solver for quadratic programs
• Zero-shot Imitation Learning from Demonstrations for Legged Robot Visual Navigation
• Statistical inference in two-sample summary-data Mendelian randomization using robust adjusted profile score
• A Statistical Learning Assessment of Huber Regression
• A Systematic Approach to Featurization for Cancer Drug Sensitivity Predictions with Deep Learning
• High Speed Tracking With A Fourier Domain Kernelized Correlation Filter
• Learning to Delegate for Large-scale Vehicle Routing
• Robust Estimation and Generative Adversarial Nets
• UPRec: User-Aware Pre-training for Recommender Systems
• Manipulating Machine Learning: Poisoning Attacks and Countermeasures for Regression Learning
• Color2Embed: Fast Exemplar-Based Image Colorization using Color Embeddings
• Model Repair: Robust Recovery of Over-Parameterized Statistical Models
• Finite-sample analysis of M-estimators using self-concordance
• Anchor regression: heterogeneous data meets causality
• Implicit Regularization in Tensor Factorization
• Real-Time User-Guided Image Colorization with Learned Deep Priors
• Relocalization, Global Optimization and Map Merging for Monocular Visual-Inertial SLAM
• A New Empirical Relation between Surface Wave Magnitude and Rupture Length for Turkey Earthquakes
• Accelerating Least Squares Imaging Using Deep Learning Techniques
• Quasi-Newton methods for large-scale electromagnetic inverse problems
• Adversarial Robust Low Rank Matrix Estimation: Compressed Sensing and Matrix Completion
• Graph-based Thermal-Inertial SLAM with Probabilistic Neural Networks
• Improved RANSAC performance using simple, iterative minimal-set solvers
• Strictly Proper Scoring Rules, Prediction, and Estimation
• Robust change point tests by bounded transformations
• Efficient Ridesharing Dispatch Using Multi-Agent Reinforcement Learning
• Optimal solutions to the isotonic regression problem
• On general notions of depth for regression
• A data driven trimming procedure for robust classification
• Learning with Fenchel-Young Losses
• Robust Sparse Reduced Rank Regression in High Dimensions
• Phase Transition Unbiased Estimation in High Dimensional Settings
• Higher Order Expansion for the MSE of M-estimators on shrinking neighborhoods
• A General Method for Robust Bayesian Modeling
• Do we need to estimate the variance in robust mean estimation?
• DORO: Distributional and Outlier Robust Optimization
• Robust estimations for the tail index of Weibull-type distribution
• Deep Residual Reinforcement Learning
• Look up and ahead: How climate scenarios affect European sovereign credit risk
• Deep Predictive Learning of Carotid Stenosis Severity
• A Generalized Framework for Edge-preserving and Structure-preserving Image Smoothing
• Statistical Inference in High-dimensional Generalized Linear Models with Streaming Data
• Patterns of Effects and Sensitivity Analysis for Differences-in-Differences
• Decoding Molecular Graph Embeddings with Reinforcement Learning
• Learning Optimal Personalized Treatment Rules Using Robust Regression Informed K-NN
• Speech gesture generation from the trimodal context of text, audio, and speaker identity
• Measuring Blood Glucose Concentrations in Photometric Glucometers\n Requiring Very Small Sample Volumes
• Maximum Mutation Reinforcement Learning for Scalable Control
• List-Decodable Mean Estimation in Nearly-PCA Time
• Escaping Poor Local Minima in Large Scale Robust Estimation
• Vector representations of text data in deep learning
• Data Dependent Convergence for Distributed Stochastic Optimization
• Robust Learning of Fixed-Structure Bayesian Networks in Nearly-Linear Time
• Learning Structured Distributions From Untrusted Batches: Faster and Simpler
• Accounting for total variation and robustness in profiling health care providers
• Pose from Shape: Deep Pose Estimation for Arbitrary 3D Objects
• Observe and Look Further: Achieving Consistent Performance on Atari
• Adaptive Capped Least Squares
• Settling the Robust Learnability of Mixtures of Gaussians
• On the power of adaptivity in statistical adversaries
• Comprehensible Context-driven Text Game Playing
• Optimal Robust Linear Regression in Nearly Linear Time
• Anatomy-Aware Cardiac Motion Estimation
• Learning Multivariate Log-concave Distributions
• Depth Completion via Deep Basis Fitting
• Using Robust Regression to Find Font Usage Trends
• Identifying Outliers using Influence Function of Multiple Kernel Canonical Correlation Analysis
• Using a Logarithmic Mapping to Enable Lower Discount Factors in\n Reinforcement Learning
• Non-decreasing Quantile Function Network with Efficient Exploration for Distributional Reinforcement Learning
• Estimating location parameters in entangled single-sample distributions
• Tilted Empirical Risk Minimization
• Improving Robustness via Risk Averse Distributional Reinforcement Learning
• Stability of SGD: Tightness Analysis and Improved Bounds
• Asymptotically Optimal Bias Reduction for Parametric Models
• DeepTIO: A Deep Thermal-Inertial Odometry with Visual Hallucination
• Robust Regression Revisited: Acceleration and Improved Estimation Rates
• Predicting Vehicles' Longitudinal Trajectories and Lane Changes on Highway On-Ramps
• Variance Breakdown of Huber (M)-estimators: n/p → m ∈ (1,∞)
• Data Science Principles for Interpretable and Explainable AI
• Adaptive Regularization of Some Inverse Problems in Image Analysis
• Neural Programmer: Inducing Latent Programs with Gradient Descent
• Robust Covariance Estimation for High-dimensional Compositional Data with Application to Microbial Communities Analysis
• Solving Irregular and Data-enriched Differential Equations using Deep\n Neural Networks
• M-Estimation Method Based Asymmetric Objective Function
• An Exploration of Deep Learning Methods in Hungry Geese
• The Flip Side of the Reweighted Coin: Duality of Adaptive Dropout and\n Regularization
• Charge-Based Prison Term Prediction with Deep Gating Network
• Tactical Optimism and Pessimism for Deep Reinforcement Learning
• Monocular visual-inertial SLAM algorithm combined with wheel speed anomaly detection
• Robust correlation analyses: false positive and power validation using a new open source matlab toolbox. [europepmc]
• Dual-energy cone-beam CT with a flat-panel detector: effect of reconstruction algorithm on material classification. [europepmc]
• Quantitative 3D-OCT motion correction with tilt and illumination correction, robust similarity measure and regularization. [europepmc]
• Controlling false discoveries in high-dimensional situations: boosting with stability selection. [europepmc]
• PredRSA: a gradient boosted regression trees approach for predicting protein solvent accessibility. [europepmc]
• Polychromatic Iterative Statistical Material Image Reconstruction for Photon-Counting Computed Tomography. [europepmc]
• MeSH Now: automatic MeSH indexing at PubMed scale via learning to rank. [europepmc]
• Immunosenescence patterns differ between populations but not between sexes in a long-lived mammal. [europepmc]
• Autistic traits, but not schizotypy, predict increased weighting of sensory information in Bayesian visual integration. [europepmc]
• Genotyping Polyploids from Messy Sequencing Data. [europepmc]
• How to Address Non-normality: A Taxonomy of Approaches, Reviewed, and Illustrated. [europepmc]
• MMSplice: modular modeling improves the predictions of genetic variant effects on splicing. [europepmc]
• Altered anticipation and processing of aversive interoceptive experience among women remitted from bulimia nervosa. [europepmc]
• X-ray imaging of a water bear offers a new look at tardigrade internal anatomy. [europepmc]
• Joint retina segmentation and classification for early glaucoma diagnosis. [europepmc]
• The Association between Serum Magnesium Levels and Depression in an Adult Primary Care Population. [europepmc]
• Optical biopsy identification and grading of gliomas using label-free visible resonance Raman spectroscopy. [europepmc]
• Environmental Risk Perception, Risk Culture, and Pro-Environmental Behavior. [europepmc]
• Effects of Hypertension, Diabetes, and Smoking on Age and Sex Prediction from Retinal Fundus Images. [europepmc]
• NEBULA is a fast negative binomial mixed model for differential or co-expression analysis of large-scale multi-subject single-cell data. [europepmc]
• Multianalyte serology in home-sampled blood enables an unbiased assessment of the immune response against SARS-CoV-2. [europepmc]
• Statistical methods for Mendelian randomization in genome-wide association studies: A review. [europepmc]
• Aging is associated with a systemic length-associated transcriptome imbalance. [europepmc]
• Proteomics and cytokine analyses distinguish myalgic encephalomyelitis/chronic fatigue syndrome cases from controls. [europepmc]
• Systematic analysis of drug combinations against Gram-positive bacteria. [europepmc]

Related

• Robust estimations from distribution structures: I. Mean
• Robust location estimation with missing data
• On Some Robust Estimates of Location
• On (in)consistency of M-estimators under contamination
• Performance of the empirical median for location estimation in heteroscedastic settings