We show that several popular few-shot learning benchmarks can be solved with varying degrees of success without using support set Labels at Test-time (LT). To this end, we introduce a new baseline called Centroid Networks, a modification of Prototypical Networks in which the support set labels are hidden from the metho…
Framework benchmarks optimizers on multiple criteria.
problem Benchmarking optimizers across diverse test functions.
method Union-free generic depth function for partial orders/rankings.
result Identifies central and outlying rankings of optimizers.
Study benchmarks TSC algorithms in distinguishing diffusions using the likelihood ratio test.
problem Benchmarking optimality of TSC algorithms in distinguishing diffusion processes.
method Proposes to benchmark TSC algorithms using the likelihood ratio test (LRT).
result LRT benchmarks are computationally efficient and can be applied to various time series types.
INT benchmark tests theorem proving agents' ability to generalize to unseen theorems.
problem Evaluating theorem proving agents' ability to generalize to unseen theorems.
method INT benchmark based on a theorem generation and proof procedure with adjustable knobs for measuring 6 types of generalization.
result MCTS can help agents prove new theorems.
Research benchmarks LLMs in medical domain to reduce hallucinations.
problem Hallucinations in medical LLMs can lead to incorrect information.
method Developed Med-HALT dataset and testing methods.
result Significant performance differences among LLMs identified.
Deep neural networks solve Raven's Progressive Matrices with high accuracy.
problem Testing relational reasoning in machine learning systems.
method Combining Wild Relation Networks with Multi-Layer Relation Networks and introducing Magnitude Encoding.
result Deep neural networks achieve 98.0 percent accuracy, significantly improving over previous methods.
Study reveals pervasive label errors in test sets, affecting machine learning benchmarks.
problem Label errors in test sets destabilize machine learning benchmarks.
method Identified label errors in 10 common datasets using confident learning algorithms and human validation.
result Lower capacity models may be more useful in real-world datasets with high proportions of erroneously labeled data.
Unified benchmarks assess data poisoning and backdoor attacks.
problem Unclear danger and effectiveness of data poisoning methods.
method Developed standardized benchmarks for data poisoning and backdoor attacks.
result Existing methods may not generalize to realistic settings.
Excessive reuse of test data has become commonplace in today's machine learning workflows. Popular benchmarks, competitions, industrial scale tuning, among other applications, all involve test data reuse beyond guidance by statistical confidence bounds. Nonetheless, recent replication studies give evidence that popular…
Temporal information impacts only a fraction of time series datasets, skewing benchmark evaluations.
problem Temporal information's impact on time series classification is often overestimated.
method Permutation tests on UCR archive to identify datasets where temporal info is irrelevant.
result Many tabular datasets perform well without temporal info, skewing benchmark evaluations.
The repeated community-wide reuse of test sets in popular benchmark problems raises doubts about the credibility of reported test-error rates. Verifying whether a learned model is overfitted to a test set is challenging as independent test sets drawn from the same data distribution are usually unavailable, while other …
Performance of investment managers are evaluated in comparison with benchmarks, such as financial indices. Due to the operational constraint that most professional databases do not track the change of constitution of benchmark portfolios, standard tests of performance suffer from the "look-ahead benchmark bias," when t…
New benchmark for non-rigid 3D human shape retrieval.
problem Distinguishing between body shapes of 3D human models.
method Extended benchmark with 145 new models and FAUST dataset.
result Improved comparison of 25 shape retrieval methods.
Benchmark tests spoken language models for infant language learning.
problem Understanding how infants learn language from speech.
method Developed a language-acquisition-friendly benchmark.
result Benchmarking shows models' strengths and weaknesses.
The paper benchmarks OS with RCTs, accounting for right-censoring.
problem Benchmarking observational studies with experimental data under censoring.
method Two cases: independent and dependent censoring. Censoring-doubly-robust signal for CATE.
result Effectiveness of censoring-aware tests verified via experiments and real data.
Study tests whether trade-off functions are above or below benchmarks using finite samples.
problem Testing trade-off functions between unknown distributions.
method Identifies a condition for nontrivial testing, constructs a test with error guarantees, and inverts the test for confidence bands.
result Finite-sample testing is possible under specific structural assumptions about rejection regions.
New metrics fail adversarial tests, with some more robust than others.
problem Evaluation metrics for time-series anomaly detection were improved but not fully robust.
method Adversarial stress-testing of 12 adopted metrics on real benchmarks.
result Some metrics are more robust than others, with ROC-based metrics being gamed more often.
We study the portfolio problem of maximizing the outperformance probability over a random benchmark through dynamic trading with a fixed initial capital. Under a general incomplete market framework, this stochastic control problem can be formulated as a composite pure hypothesis testing problem. We analyze the connecti…
Benchmarking recursive collapse claims with a new framework under false-positive control.
problem Evaluating recursive systems for failure patterns and warning claims.
method Developed Loopzero framework for testing recursive failures, specified claim boundaries in Lean, evaluated under FP constraint, and compared with standard detectors.
result No standard detectors or Loopzero's pre-registered quantile detector achieved the required operating point under the false-positive contract.
This work benchmarks diffusion model-based samplers for Bayesian inverse problems.
problem Optimizing diffusion models for uncertainty quantification in Bayesian inverse problems.
method Introduces three benchmark problems and a unified framework for diffusion model-based posterior sampling.
result Provides insights into strengths and limitations of diffusion model-based samplers.
Study benchmarks classical models over quantum in DeFi yield prediction.
problem Accurate yield and performance forecasting for DeFi liquidity allocation.
method Benchmarked six models on Curve Finance pools' historical data.
result Classical models, especially XGBoost, outperform quantum models.
While current benchmark reinforcement learning (RL) tasks have been useful to drive progress in the field, they are in many ways poor substitutes for learning with real-world data. By testing increasingly complex RL algorithms on low-complexity simulation environments, we often end up with brittle RL policies that gene…
We develop a statistical framework to benchmark and select large language models based on their risks.
problem Benchmarking and selecting large language models based on their associated risks.
method A distributional framework using first and second order stochastic dominance, linked to mean-risk models in finance.
result Formalizes a risk-aware approach for model selection, balancing risk and utility.
SD-SCMs generate counterfactual data for causal inference benchmarks.
problem Benchmarking causal inference methods with realistic data.
method Sequence-driven structural causal models (SD-SCMs) for causal inference.
result State-of-the-art methods struggle with individual treatment effect estimation.
We present Fashion-MNIST, a new dataset comprising of 28x28 grayscale images of 70,000 fashion products from 10 categories, with 7,000 images per category. The training set has 60,000 images and the test set has 10,000 images. Fashion-MNIST is intended to serve as a direct drop-in replacement for the original MNIST dat…
Microdata improves inflation forecasts after major shocks, study finds.
problem Forecasting inflation in a non-stationary environment with microeconomic data.
method Developed a scan test to detect periods of micro forecast outperformance, combined with adaptive machine learning.
result Micro forecasts improve inflation predictions after major shocks, especially after 2020.
Study tests five popular trading signal families and finds four refuted, one inconclusive, and one not refuted.
problem Testing the viability of five popular trading signal families for generating a positive edge.
method Statistical edge testing, economic viability assessment, and finite-bankroll survival under leverage using exposure-matched benchmarks, stationary-bootstrap confidence intervals, and hierarchical Benjamini-Yekutieli control.
result Four out of five signal families are refuted, one is inconclusive, and one is not refuted.
It is common that a trained classification model is applied to the operating data that is deviated from the training data because of noise. This paper demonstrates that an ensemble classifier, Diversified Multiple Tree (DMT), is more robust in classifying noisy data than other widely used ensemble methods. DMT is teste…
A wild bootstrap method for nonparametric hypothesis tests based on kernel distribution embeddings is proposed. This bootstrap method is used to construct provably consistent tests that apply to random processes, for which the naive permutation-based bootstrap fails. It applies to a large group of kernel tests based on…
In this paper, we propose Test-Time Training, a general approach for improving the performance of predictive models when training and test data come from different distributions. We turn a single unlabeled test sample into a self-supervised learning problem, on which we update the model parameters before making a predi…
Novel strategy benchmarks observational studies against randomized trials.
problem Benchmarking observational studies for treatment effect bias.
method Statistical test for null hypothesis of treatment effect difference.
result Valid lower bound on maximum bias strength for any subgroup.
New benchmarks focus on LLM risk in finance, not just accuracy.
problem Standard benchmarks ignore LLM financial risks, leading to unsafe deployment.
method Three-level agenda: model, workflow, and system stress-testing.
result Hidden weaknesses in LLMs are missed by traditional benchmarks.
Paper compiles ML algorithm performance benchmarks on OpenML datasets.
problem Finding optimal hyperparameters for ML algorithms efficiently.
method Generated benchmark data for 7 ML algorithms on 39 datasets, fixed hyperparameters before testing.
result Comprehensive dataset of ML algorithm performance sensitivity.
Recently many efforts have been made to incorporate persistence diagrams, one of the major tools in topological data analysis (TDA), into machine learning pipelines. To better understand the power and limitation of persistence diagrams, we carry out a range of experiments on both graph data and shape data, aiming to de…
Tiny benchmarks reduce LLM evaluation costs by using fewer examples.
problem Expensive evaluation of LLMs with tens of thousands of examples.
method Developed evaluation tools and tiny versions of popular benchmarks.
result Accurately estimate LLM performance with just 100 curated examples.
New method assesses multivariate stochastic dominance using Optimal Transport.
problem Benchmarking models across multiple metrics considering dependencies.
method Characterization of multivariate first stochastic dominance via couplings, entropic regularization, and Optimal Transport.
result Established CLT and consistency for the empirical statistic, enabling hypothesis testing.
In this paper, we investigate the problem of overfitting in deep reinforcement learning. Among the most common benchmarks in RL, it is customary to use the same environments for both training and testing. This practice offers relatively little insight into an agent's ability to generalize. We address this issue by usin…
Unified Python package N2 benchmarks NN-based matrix completion methods.
problem Matrix completion with robust NN methods across diverse applications.
method Unified Python package N2 consolidates NN-based methods. result NN-based techniques outperform classical methods in real-world settings.
New method compares classifiers using GSD-front, addressing statistical uncertainty and robustness.
problem Comparing classifiers with multiple quality metrics and statistical uncertainty.
method Proposes GSD-front and statistical tests for robust comparisons.
result Reliable method for comparing classifiers with statistical uncertainty and robustness.
Real-world machine learning applications often have complex test metrics, and may have training and test data that are not identically distributed. Motivated by known connections between complex test metrics and cost-weighted learning, we propose addressing these issues by using a weighted loss function with a standard…
We build new test sets for the CIFAR-10 and ImageNet datasets. Both benchmarks have been the focus of intense research for almost a decade, raising the danger of overfitting to excessively re-used test sets. By closely following the original dataset creation processes, we test to what extent current classification mode…
Two semimetrics on probability distributions are proposed, given as the sum of differences of expectations of analytic functions evaluated at spatial or frequency locations (i.e, features). The features are chosen so as to maximize the distinguishability of the distributions, by optimizing a lower bound on test power f…
Paper proposes a new speech representation benchmark and model.
problem Lack of benchmarks for comparing speech representations.
method Unsupervised triplet-loss objective for training a universal non-semantic speech representation.
result Proposed representation outperforms other models on benchmark and transfer learning tasks.
AutoML simplifies two-sample tests for detecting distribution shifts.
problem Detecting distribution shifts between datasets.
method Uses mean discrepancy of a witness function with squared loss minimization.
result AutoML simplifies and improves two-sample testing performance.
BREEDS benchmarks assess model robustness to subpopulation shifts.
problem Measuring model robustness to novel subpopulation shifts.
method Controlled synthesis of realistic distribution shifts using class structure.
result Validated model sensitivity and effectiveness of robustness interventions.
Adaptive compute allocation improves model performance by prioritizing harder queries.
problem Inefficiency in allocating test-time compute uniformly across all queries.
method Formulated as a bandit learning problem, proposed adaptive algorithms that estimate query difficulty and allocate compute accordingly.
result Achieved up to 15.29% relative performance improvement on various benchmarks.
Hypothesis testing is an important cognitive process that supports human reasoning. In this paper, we introduce a computational hypothesis testing approach based on memory augmented neural networks. Our approach involves a hypothesis testing loop that reconsiders and progressively refines a previously formed hypothesis…
Deployment-complete benchmarking assesses if evidence leads to consistent deployment actions.
problem Lack of clear evidence leading to consistent deployment actions.
method Introduces deployment-complete benchmarking to test if benchmark evidence determines deployment actions.
result Benchmark evidence must be complete for a claim to lead to a consistent deployment action.