The paper introduces sanity tests to detect spurious correlations in AI-guided radiology systems.
problem Detecting when AI systems perform well on development data for the wrong reasons.
method Design and implementation of sanity tests to identify spurious correlations.
result Sanity tests can identify spurious correlations in AI-guided radiology systems.
Randomly initialized networks can perform as well as pruned networks.
problem Understanding and improving network pruning methods.
method Sanity checks on recent pruning methods, proposing random tickets.
result Randomly initialized networks can perform as well as pruned networks.
There is great interest in "saliency methods" (also called "attribution methods"), which give "explanations" for a deep net's decision, by assigning a "score" to each feature/pixel in the input. Their design usually involves credit-assignment via the gradient of the output with respect to input. Recently Adebayo et al.…
Reliability is a critical consideration to DL-based systems. But the statistical nature of DL makes it quite vulnerable to invalid inputs, i.e., those cases that are not considered in the training phase of a DL model. This paper proposes to perform data sanity check to identify invalid inputs, so as to enhance the reli…
Study tests if input gradients highlight discriminative features, finds they often fail.
problem Validity of assumption that input gradients highlight discriminative features in model predictions.
method Developed DiffROAR framework and BlockMNIST dataset to test assumption on four benchmarks.
result Input gradients of standard models often fail to highlight discriminative features, while robust models do.
Two modified tests improve the reliability of evaluating explanation methods.
problem Methodological concerns in evaluating explanation methods for saliency maps.
method Proposed modifications to the Model Parameter Randomisation Test (MPRT): Smooth MPRT and Efficient MPRT.
result Enhanced metric reliability, facilitating more trustworthy deployment of explanation methods.
Differentially private learning on real-world data poses challenges for standard machine learning practice: privacy guarantees are difficult to interpret, hyperparameter tuning on private data reduces the privacy budget, and ad-hoc privacy attacks are often required to test model privacy. We introduce three tools to ma…
The ability to perform effective off-policy learning would revolutionize the process of building better interactive systems, such as search engines and recommendation systems for e-commerce, computational advertising and news. Recent approaches for off-policy evaluation and learning in these settings appear promising. …
The paper tackles confidence calibration for exploratory machine learning problems.
problem Difficulty in curating datasets and confusion about category validity.
method Introduces four new algorithms for category-specific confidence estimation, including kernel density ratios.
result Kernel density ratios provide a novel approach to confidence calibration, especially for exploratory problems.
Paper presents a universal baseline for binary prediction models.
problem Need a robust baseline to evaluate model performance.
method Dutch Draw (DD) baseline method for binary classification models.
result Reduces to almost always predicting zero or one in most situations.
Detecting adversarial examples is as hard as classifying them.
problem The difficulty of detecting adversarial examples in machine learning models.
method Proved a general hardness reduction between detection and classification of adversarial examples.
result The hardness reduction implies that detecting adversarial examples is computationally infeasible.
The study evaluates saliency metrics for image classifier outputs, finding inconsistencies and unreliability.
problem Inconsistencies and unreliability in saliency metrics for evaluating pixel relevance.
method Investigated existing saliency metrics, calculated and compared their consistency, and applied psychometric testing methods.
result Saliency metrics can be statistically unreliable and inconsistent, affecting comparative rankings.
In this article, we derive concentration inequalities for the cross-validation estimate of the generalization error for empirical risk minimizers. In the general setting, we prove sanity-check bounds in the spirit of \cite{KR99} \textquotedblleft\textit{bounds showing that the worst-case error of this estimate is not m…
Saliency methods have emerged as a popular tool to highlight features in an input deemed relevant for the prediction of a learned model. Several saliency methods have been proposed, often guided by visual appeal on image data. In this work, we propose an actionable methodology to evaluate what kinds of explanations a g…
Scientific discovery is limited by hypothesis redundancy, and hybrid methods can exploit non-local exploration.
problem Limitation of scientific discovery due to hypothesis redundancy.
method Hybrid discovery systems combining structured local search with LLM-generated non-local proposals.
result Hybrid methods can exploit non-local exploration when three geometric conditions co-occur.
Saliency methods can aid understanding of deep neural networks. Recent years have witnessed many improvements to saliency methods, as well as new ways for evaluating them. In this paper, we 1) present a novel region-based attribution method, XRAI, that builds upon integrated gradients (Sundararajan et al. 2017), 2) int…
This study optimizes trading strategy parameters using walk-forward techniques and finds robust performance.
problem Optimizing trading strategy performance through parameter optimization.
method Walk-forward optimization with varying window lengths, tested on Bitcoin, Binance Coin, and Ethereum.
result The strategy outperforms Buy-and-Hold with lower drawdown and higher Information Ratio.
In neural networks, it is often desirable to work with various representations of the same space. For example, 3D rotations can be represented with quaternions or Euler angles. In this paper, we advance a definition of a continuous representation, which can be helpful for training deep neural networks. We relate this t…
FDN improves probabilistic regressors' adaptability to distribution shifts.
problem Overconfidence in modern probabilistic regressors under distribution shift.
method FDN uses input-conditioned distributions over network weights, trained with a Monte Carlo beta-ELBO objective.
result FDN produces predictive mixtures whose dispersion adapts to the input, providing shift-aware uncertainty.
Proposes a copula-based filter for diabetes risk prediction.
problem Feature selection for robust and interpretable predictive modeling in medicine, especially for extreme patient strata.
method Copula-based supervised filter using Gumbel-copula implied upper-tail concordance score (lambda U).
result The proposed filter outperforms standard filters and provides clinically coherent predictors.
We propose a neural network approach to price EU call options that significantly outperforms some existing pricing models and comes with guarantees that its predictions are economically reasonable. To achieve this, we introduce a class of gated neural networks that automatically learn to divide-and-conquer the problem …
Probability density estimation is a classical and well studied problem, but standard density estimation methods have historically lacked the power to model complex and high-dimensional image distributions. More recent generative models leverage the power of neural networks to implicitly learn and represent probability …
Recently, several methods have been proposed to explain the predictions of recurrent neural networks (RNNs), in particular of LSTMs. The goal of these methods is to understand the network's decisions by assigning to each input variable, e.g., a word, a relevance indicating to which extent it contributed to a particular…
Explaining the output of a complicated machine learning model like a deep neural network (DNN) is a central challenge in machine learning. Several proposed local explanation methods address this issue by identifying what dimensions of a single input are most responsible for a DNN's output. The goal of this work is to a…
New method interprets deep neural networks for better recommendation system understanding.
problem Making deep neural networks explainable for better user trust and understanding.
method Proposes a novel formulation of interpretable deep neural networks using masked weights and hidden features.
result Demonstrates models achieving close predictive performance with informative attributions.
We aim to create the highest possible quality of treatment-control matches for categorical data in the potential outcomes framework. Matching methods are heavily used in the social sciences due to their interpretability, but most matching methods do not pass basic sanity checks: they fail when irrelevant variables are …
Large dataset released for ITE and UM research.
problem Estimating causal impact of actions in various sectors.
method Release of a large dataset, formalization of UM, synthetic response surfaces, heterogeneous treatment assignment.
result Validation of ITE prediction and UM methods with high statistical significance.
Study examines XAI methods for ECG analysis to improve model transparency.
problem Lack of transparency in deep learning models for ECG analysis.
method Investigates post-hoc XAI methods for local and global perspectives, establishes sanity checks, and demonstrates knowledge discovery.
result Quantitative evidence supports expert rules for sensible attribution methods and demonstrates XAI's utility for knowledge discovery.
Study on sensor fusion algorithms under high dimensional noise.
problem Behavior of sensor fusion algorithms under high dimensional noise.
method Analysis of NCCA and AD algorithms using Gaussian kernel.
result Robustness of NCCA and AD to high dimensional noise depends on SNR and bandwidth selection.
Study reveals biases in ImageNet models are not sufficient for generalization.
problem Understanding and improving generalization of image classification models.
method Large-scale study of 48 ImageNet models trained via various methods.
result Biases identified in ImageNet models do not fully explain generalization.
Deep learning techniques have proven high accuracy for identifying melanoma in digitised dermoscopic images. A strength is that these methods are not constrained by features that are pre-defined by human semantics. A down-side is that it is difficult to understand the rationale of the model predictions and to identify …
Proposes MEED framework for model interpretation.
problem Improving model interpretability and avoiding undesired characteristics.
method Adversarial Infidelity Learning (AIL) for effective feature selection.
result AIL mechanism helps learn desired conditional distribution.
RayS attack improves hard-label adversarial attacks by reducing query complexity and identifying false robust models.
problem Challenges in hard-label adversarial attacks, especially in terms of effectiveness and efficiency.
method Reformulates continuous problem into discrete problem without gradient estimation and uses a fast check step to eliminate unnecessary searches.
result Significantly reduces the number of queries needed for hard-label attacks and identifies false robust models.
LLM forecasting benchmarks suffer from information leakage, which confounds model performance.
problem LLM forecasting benchmarks suffer from information leakage.
method A retrieval-augmented LLM forecaster observes only decision-time information.
result The full pipeline obtains a median monthly Spearman rank IC of +0.154.
Study of Polymarket's prediction market microstructure using tick-level order book data.
problem Understanding the microstructure of decentralized prediction markets.
method Analysis of a continuous tick-level order book feed and on-chain trade records.
result Trade direction inferred from Polymarket's public order-book feed disagrees with on-chain data in ~59% of cases.
Unified analysis of multilabel Fisher discriminants with improved dimensionality and robustness.
problem Improving discriminant analysis for multilabel classification with enhanced dimensionality and robustness.
method Unified theoretical analysis of multilabel Fisher discriminants with algebraic and statistical guarantees.
result Unified characterization of multilabel Fisher objectives and their equivalence under orthogonality constraints.
Unified analysis of multilabel Fisher discriminants with improved dimensionality and robustness.
problem Improving discriminant analysis for multilabel classification with enhanced dimensionality and robustness.
method Unified algebraic and statistical analysis of multilabel Fisher discriminants with Stiefel orthogonality constraints.
result Equivalence of four Fisher objectives under the Stiefel constraint and improved discriminant dimensionality.
Many methods for automated software test generation, including some that explicitly use machine learning (and some that use ML more broadly conceived) derive new tests from existing tests (often referred to as seeds). Often, the seed tests from which new tests are derived are manually constructed, or at least simpler t…
A family of maximum mean discrepancy (MMD) kernel two-sample tests is introduced. Members of the test family are called Block-tests or B-tests, since the test statistic is an average over MMDs computed on subsets of the samples. The choice of block size allows control over the tradeoff between test power and computatio…
USP test improves on Pearson's chi-squared and G-test for independence.
problem Deficiencies in Pearson's chi-squared and G-test for independence. method USP test based on U-statistic estimator of population dependence measure. result USP test controls size, handles small cell counts, and detects minimal violations of independence.
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.
This paper provides a comprehensive survey of Machine Learning Testing (ML testing) research. It covers 144 papers on testing properties (e.g., correctness, robustness, and fairness), testing components (e.g., the data, learning program, and framework), testing workflow (e.g., test generation and test evaluation), and …
Cheap permutation tests speed up distribution testing without sacrificing accuracy.
problem Efficiently testing distribution differences and independence.
method Group datapoints into bins and permute only these bins, using stored sufficient statistics.
result Cheap permutation tests maintain the accuracy and optimality of standard tests but are significantly faster.
Paper proposes a chi-square test for distance correlation.
problem Testing distance correlation is computationally expensive.
method Proposes a chi-square test for distance correlation, non-parametric, fast, applicable to various metrics.
result Chi-square test exhibits similar power to permutation test and can be valid and universally consistent for testing independence.
The paper tests properties of multiple distributions with limited samples.
problem Testing properties of multiple distributions with few samples.
method Designing testers for uniformity, identity, and closeness testing under specific conditions.
result Sample optimal testers for uniformity, identity, and closeness testing are provided.
Optimizes two-sample tests for non-Euclidean domains using spectral regularization.
problem Optimizing two-sample tests for non-Euclidean domains.
method Spectral regularization of MMD test to achieve minimax optimality.
result Proposes a spectral regularization method that improves test optimality.
DRIFT uses RL to automate functional software testing efficiently.
problem Efficient and reliable automated software testing.
method DRIFT employs Q-learning with Graph Neural Networks on symbolic UI representations.
result DRIFT can robustly test software functionalities in a fully automated manner.
Post hoc test for Sharpe ratio improves pairwise comparisons.
problem Improving pairwise comparisons of Sharpe ratios.
method Analogous to Tukey's test, applied after rejecting equal Signal-Noise ratios.
result Maintains nominal type I rate and is moderately powerful.