Unified framework for calculating Shapley values with correlated features.
problem Inaccurate Shapley values due to multicollinearity.
method Unified framework with matrix formulation and mathematical proof.
result Shapley values become independent of feature correlations.
Paper finds Dutch Draw optimal baseline for binary classification.
problem Need a proper baseline for binary classification validation.
method Examined all input-independent baseline methods.
result Dutch Draw is optimal baseline under given conditions.
CIRCE measures conditional independence for learning invariant features.
problem Learning invariant features while being conditionally independent of a distractor.
method CIRCE is a measure of conditional independence applied as a regularizer in feature learning.
result CIRCE provides a zero value if and only if features are conditionally independent of the distractor given the target.
New methods using vine copulas improve accuracy of feature dependence in predictive models.
problem Inaccurate feature dependence assumptions in Shapley values lead to incorrect explanations.
method Proposed two new approaches based on vine copulas to model feature dependence.
result Vine copula approaches give more accurate approximations to true Shapley values.
Efficiently estimates SAGE values using causal structure learning.
problem Computational infeasibility of exact SAGE calculations.
method Uses causal structure learning to identify conditional independencies and accelerate SAGE approximation.
result Empirically demonstrates efficient and accurate estimation of SAGE values.
The goal of supervised feature selection is to find a subset of input features that are responsible for predicting output values. The least absolute shrinkage and selection operator (Lasso) allows computationally efficient feature selection based on linear dependency between input features and output values. In this pa…
ECCIT improves conditional independence tests by calibrating for miscalibration.
problem Inaccurate frequentist guarantees in CITs, especially in small samples and misspecified models.
method Empirically Calibrated Conditional Independence Tests (ECCIT) that optimize and correct for miscalibration.
result ECCIT achieves valid FDR with higher power than existing calibration strategies.
Shapley value improves model interpretation but not causal inference.
problem Improving model interpretability without losing predictive power.
method Analyzed Shapley value in Bayesian networks, linking it to conditional independence.
result Eliminating high Shapley value variables does not harm predictive performance, but low Shapley value variables can.
New Shapley values reveal non-linear feature dependencies.
problem Understanding non-linear dependencies in machine learning models.
method Model-independent Shapley values using non-parametric measures of dependence.
result Model-independent Shapley values can uncover non-linear dependencies.
Conditional independence testing is a key problem required by many machine learning and statistics tools. In particular, it is one way of evaluating the usefulness of some features on a supervised prediction problem. We propose a novel conditional independence test in a predictive setting, and show that it achieves bet…
Study shows safely discarding features based on aggregate SHAP values is sound.
problem Discarding features based on aggregate SHAP values without proper justification.
method Investigated the soundness of discarding features based on aggregate SHAP values, proposing to aggregate SHAP values over the extended support.
result A small aggregate SHAP value implies safely discarding the corresponding feature.
New versions of the set-valued average value at risk for multivariate risks are introduced by generalizing the well-known certainty equivalent representation to the set-valued case. The first "regulator" version is independent from any market model whereas the second version, called the market extension, takes trading …
Explaining complex or seemingly simple machine learning models is an important practical problem. We want to explain individual predictions from a complex machine learning model by learning simple, interpretable explanations. Shapley values is a game theoretic concept that can be used for this purpose. The Shapley valu…
Sampling strategies significantly affect feature approximations in ELA, impacting classifier accuracy.
problem The impact of sampling strategies on feature approximations in ELA.
method Analysis of feature approximations from different sampling strategies and sample sizes.
result Feature approximations from different sampling strategies do not converge, affecting classifier accuracy.
We adapt Shapley values to explain model uncertainty, connecting it to information theory.
problem Explaining uncertainty in model predictions.
method Adapted Shapley value framework to quantify feature contributions to predictive uncertainty.
result Deep connections between Shapley values and information theory quantities.
New measure of feature influence in classification problems considering feature dependencies.
problem Measuring the influence of features in classification problems with dependencies.
method Developed a new measure based on cooperative game theory, providing axiomatic characterization and demonstrating its equivalence to the Banzhaf-Owen value.
result The proposed influence measure effectively characterizes feature importance in classification problems with feature dependencies.
A new method explains mixed features for predictive models using conditional inference trees.
problem Explaining complex machine learning models with mixed features.
method Proposes a method to explain mixed features (continuous, discrete, ordinal, categorical) using conditional inference trees.
result Our method often outperforms current industry standards in various simulation studies and real-world financial data.
We solve the ANOVA decomposition for categorical inputs.
problem Lack of a closed-form expression for ANOVA decomposition with categorical dependent variables.
method Bridge functional analysis with discrete Fourier analysis to derive a closed-form decomposition.
result Closed-form decomposition for categorical inputs without assumptions.
Subjective classification of galaxies can mislead us in the quest of the origin regarding formation and evolution of galaxies since this is necessarily limited to a few features. The human mind is not able to apprehend the complex correlations in a manyfold parameter space, and multivariate analyses are the best tools …
This paper shows how deep neural networks can learn rich, independent features that significantly deviate from initialization.
problem Understanding how deep neural networks achieve meaningful feature learning and global convergence.
method Investigation of infinitely wide, L-layer neural networks using the tensor program framework under Maximal Update parametrization. result SGD enables these networks to learn linearly independent features that substantially deviate from their initial values, capturing relevant data information.
Efficiently plans large MDPs with weak function approximations.
problem Planning in large MDPs with limited function approximation capabilities.
method Uses linear value function approximation with weak requirements and a generative oracle.
result Produces almost-optimal actions for any state with polynomial computation time.
We propose a flexible method for estimating value functions in reinforcement learning without parametric assumptions.
problem Lack of interpretability in reinforcement learning models, especially in healthcare applications.
method Nonparametric additive model using local kernel regression and basis expansion.
result Personalized, adaptive recommendations for postoperative recovery.
New method R-LOCO improves local feature importance analysis.
problem Local attribution methods fail to accurately identify important features.
method R-LOCO segments input space into regions and applies global methods within.
result R-LOCO delivers more accurate local attributions.
Constraint-based causal discovery (CCD) algorithms require fast and accurate conditional independence (CI) testing. The Kernel Conditional Independence Test (KCIT) is currently one of the most popular CI tests in the non-parametric setting, but many investigators cannot use KCIT with large datasets because the test sca…
Feature selection methods are usually evaluated by wrapping specific classifiers and datasets in the evaluation process, resulting very often in unfair comparisons between methods. In this work, we develop a theoretical framework that allows obtaining the true feature ordering of two-dimensional sequential forward feat…
Unified feature importance for machine learning models tackles sufficiency and necessity limitations.
problem Insufficient and incomplete explanations of machine learning models.
method Formalized sufficiency and necessity notions, proposing a unified importance measure.
result Unified importance measure detects features missed by sufficiency and necessity alone.
RECol generates error columns to improve outlier detection.
problem Outlier detection in data with complex relationships.
method Generates reconstruction error columns for leave-one-out feature sets.
result Improves ROC-AUC and PR-AUC values of common outlier detection methods.
CIBP models feature abundance in latent feature models.
problem Modeling feature abundance in latent feature models.
method Proposes a new Bayesian nonparametric prior, the CIBP, for latent feature models.
result The expected number of features is bounded even as the number of objects increases.
CatNet controls FDR in LSTM models using SHAP feature importance and Gaussian mirrors.
problem Controlling False Discovery Rate (FDR) in LSTM models with feature selection.
method CatNet uses SHAP values for feature importance and Gaussian Mirror algorithm for FDR control. It introduces a kernel-based independence measure to handle feature correlations.
result CatNet reduces overfitting and improves model interpretability on simulated and real-world data.
New framework detects outliers in non-IID categorical data.
problem Existing outlier detection methods fail in non-IID data.
method Value-value graph-based representation and outlierness propagation.
result Significant improvement in AUC on complex data sets.
Proves a conjecture for Calabi-Yau manifolds.
problem Maximal degeneration of Calabi-Yau manifolds.
method Valuative independence condition for section ring.
result Metric SYZ conjecture proven.
We show that the error probability of reconstructing kernel matrices from Random Fourier Features for the Gaussian kernel function is at most O(R2/3exp(−D)), where D is the number of random features and R is the diameter of the data domain. We also provide an information-theoretic method-independen…
For various applications, the relations between the dependent and independent variables are highly nonlinear. Consequently, for large scale complex problems, neural networks and regression trees are commonly preferred over linear models such as Lasso. This work proposes learning the feature nonlinearities by binning fe…
Constructs independent bases for cubic curve families using Hessian structures.
problem Finding independent bases for cubic curve families.
method Uses a Hessian structure to define a cost function for constructing bases.
result Constructs valuatively independent bases for H0(X,Lk). E-C2ST uses E-values for high-dimensional data two-sample tests.
problem Statistical testing for high-dimensional data.
method Combines split likelihood ratio tests and predictive independence tests, using E-values for anytime-valid sequential tests.
result E-C2ST achieves enhanced statistical power by partitioning datasets into multiple batches.
The Support Vector Machine (SVM) is one of the most widely used classification methods. In this paper, we consider the soft-margin SVM used on data points with independent features, where the sample size n and the feature dimension p grows to ∞ in a fixed ratio p/n→δ. We propose a set of equatio…
New method tests conditional independence using spectral representations.
problem Untestable conditional independence in many settings.
method Spectral representations of partial covariance operators, bi-level contrastive learning.
result Asymptotic validity and power guarantees for CI testing.
We present a general framework for hypothesis testing on distributions of sets of individual examples. Sets may represent many common data sources such as groups of observations in time series, collections of words in text or a batch of images of a given phenomenon. This observation pattern, however, differs from the c…
Constructs unique bases for CY varieties over valued fields.
problem Finding unique bases for CY varieties over valued fields.
method Uses techniques from higher rank degenerations in K-stability.
result Induces canonical functions on skeletons and agrees with tropicalizations of theta functions.
We consider a firm that sells a large number of products to its customers in an online fashion. Each product is described by a high dimensional feature vector, and the market value of a product is assumed to be linear in the values of its features. Parameters of the valuation model are unknown and can change over time.…
A new feature selection method using attention for neural networks.
problem Feature selection for neural networks with a budget constraint.
method Sequential Attention: greedy forward selection with attention weights.
result Achieves state-of-the-art results for neural networks.
Introduces joint Shapley values to measure feature importance in models.
problem Measuring the importance of feature sets in machine learning models.
method Extends Shapley's axioms to measure a set of features' average contribution to a model's prediction.
result Joint Shapley values provide unique insights and are more consistent with local intuitions.
Greedy selection works well in a toy model of independent increments.
problem Iterative selection of maximum-value processes from i.i.d. stochastic processes.
method Fixed greedy selection at each stage.
result Optimal strategy is greedy selection under independent increments.
Optimal Bayesian feature filtering (OBF) is a supervised screening method designed for biomarker discovery. In this article, we prove two major theoretical properties of OBF. First, optimal Bayesian feature selection under a general family of Bayesian models reduces to filtering if and only if the underlying Bayesian m…
New features generated from kernel methods are minimally dependent on sensitive features.
problem Generating fair features in the presence of sensitive and non-sensitive features.
method Relaxed Maximum Mean Discrepancy criterion, Hilbert-space-valued conditional expectation, plug-in approach.
result Closed-form solution for minimizing dependencies between new and sensitive features.
A new computationally efficient dependence measure, and an adaptive statistical test of independence, are proposed. The dependence measure is the difference between analytic embeddings of the joint distribution and the product of the marginals, evaluated at a finite set of locations (features). These features are chose…
We present the Parallel, Forward-Backward with Pruning (PFBP) algorithm for feature selection (FS) in Big Data settings (high dimensionality and/or sample size). To tackle the challenges of Big Data FS PFBP partitions the data matrix both in terms of rows (samples, training examples) as well as columns (features). By e…
The goal of feature selection is to identify important features that are relevant to explain an outcome variable. Most of the work in this domain has focused on identifying globally relevant features, which are features that are related to the outcome using evidence across the entire dataset. We study a more fine-grain…