GGA improves untrustworthy prediction detection in neural networks without retraining.
problem Susceptibility of neural networks to untrustworthy predictions, especially adversarial attacks and out-of-distribution data.
method Geometric Gradient Analysis (GGA) analyzes the geometry of neural network loss landscapes based on saliency maps.
result GGA outperforms existing methods in detecting untrustworthy predictions, including adversarial and out-of-distribution data.
Despite widespread adoption, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust, which is fundamental if one plans to take action based on a prediction, or when choosing whether to deploy a new model. Such understanding also pr…
Bayesian framework improves survival prediction accuracy and uncertainty quantification.
problem Inaccurate uncertainty estimates in survival models.
method Bayesian framework combining variational inference, neural multi-task logistic regression, and sparsity-inducing prior.
result Better quantification of survival uncertainty and more accurate predictions.
New algorithms test independence with fewer samples by using predictive information.
problem Testing independence of distributions with limited samples.
method Augmented distribution testing framework that incorporates predictive information.
result Optimal sample complexity achieved, matching lower bounds.
SHAP clustering explains model predictions by grouping similar feature contributions.
problem Lack of explainability in black-box models.
method SHAP values for feature contributions, supervised clustering of SHAP values.
result Insight into pathways leading to similar predictions.
Rating platforms enable large-scale collection of user opinion about items (products, other users, etc.). However, many untrustworthy users give fraudulent ratings for excessive monetary gains. In the paper, we present FairJudge, a system to identify such fraudulent users. We propose three metrics: (i) the fairness of …
Optimal estimator for discrete distributions from faulty batches.
problem Estimating discrete distributions from batches, some of which may be unreliable.
method First polynomial-time estimator achieving optimal accuracy in number of batches.
result Optimal estimation accuracy in polynomial time.
Proposes a simple solution to Gini importance bias in random forests.
problem Gini importance measure in random forests is biased and unreliable.
method Computes loss reduction on out-of-bag samples instead of in-bag.
result Solves the misleading/untrustworthy Gini importance issue.
Algorithmic stablecoins optimize monetary policy to balance price stability.
problem Persistent inflation from centralized monetary policy.
method Propose and study a rule-based monetary policy model for algorithmic stablecoins.
result Optimal trade-off between price stability and supply stability.
A new model uses neural networks to efficiently learn multivariate temporal point processes.
problem Efficiently modeling multivariate temporal point processes with low parameter complexity.
method Modeling the cumulative hazard function with neural networks for each variate.
result The proposed model achieves state-of-the-art performance on data fitting and event prediction tasks.
PRESTO maps latent representations across diverse ML models.
problem Understanding variability in latent representations across different ML models.
method Uses persistent homology to characterize latent spaces and measure their pairwise similarity.
result Preserves desirable properties and enables sensitivity analysis of latent representations.
A new OOD detection method OTOD uses optimal transport theory to improve model performance.
problem Detecting unknown samples in real-world machine learning models.
method OTOD uses optimal transport theory to calculate an OOD score combining features, logits, and softmax probability space.
result OTOD outperforms state-of-the-art methods by significant margins on benchmarks.
Thanks to the advances in machine learning, data-driven analysis tools have become valuable solutions for various applications. However, there still remain essential challenges to develop effective data-driven methods because of the need to acquire a large amount of data and to have sufficient computing power to handle…
Paper presents privacy-preserving techniques for HD computing.
problem Privacy loss in HD computing due to reversible computation.
method Quantization and pruning of hypervectors for differential privacy.
result Differentially private HD model for cloud inference.
Survey on causal interpretability models for machine learning.
problem Understanding and explaining the decisions made by machine learning models.
method Comprehensive survey on causal interpretable models.
result Evaluation metrics for interpretability are discussed.
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.
Proposes a method to improve SBI under model misspecification.
problem Unreliable inference from SBI methods under model misspecification.
method Introduces a regularized loss function to penalize statistics that increase model-data mismatch.
result Demonstrates superior performance and robust inference in misspecified scenarios.
OpinionRank uses graph-based ranking to improve unreliable crowdsourced labels.
problem Improving trustworthiness of crowdsourced labels for machine learning.
method Graph-based spectral ranking to integrate unreliable labels.
result OpinionRank outperforms conventional algorithms in reliability and scalability.
New method improves Bayesian inference for parametric models, robust to misspecification.
problem Inference can be untrustworthy when parametric models are wrong.
method Adaptive nonparametric corrections for parametric Bayesian models using generalized Bayes.
result The method achieves robustness and efficiency, converging fast when the parametric model is close to true.
While machine learning (ML) models are being increasingly trusted to make decisions in different and varying areas, the safety of systems using such models has become an increasing concern. In particular, ML models are often trained on data from potentially untrustworthy sources, providing adversaries with the opportun…
Local differential privacy method protects image privacy in machine learning.
problem Protecting image privacy in machine learning models from data owners.
method Locally perturbs image representations before releasing them to the data user.
result Proposes DCAConv, an image feature extractor that maintains high data utility while preserving privacy.
PRUDEX-Compass evaluates FinRL methods on 6 axes for financial market investments.
problem Insufficient evaluation of FinRL methods in financial markets.
method Introduces PRUDEX-Compass with 6 axes and 17 measures for evaluation.
result Demonstrates the effectiveness of PRUDEX-Compass on 4 real-world datasets.
Paper addresses reward hacking in preference optimization, proposing POWER-DL to improve AI alignment.
problem Reward hacking problem in preference optimization, leading to undesired behaviors.
method POWER-DL combines robust reward maximization and dynamic label updates to mitigate reward hacking.
result POWER-DL consistently outperforms state-of-the-art methods on alignment benchmarks.
New algorithms learn from untrusted batches with improved efficiency.
problem Learning from untrusted batches with adversarial responses.
method Sum-of-Squares hierarchy applied to robust mean estimation.
result Reduces sample complexity to polylogarithmic in n for most natural distributions. Fuzzy prediction sets generalize binary predictions to include elements at varying confidence levels.
problem Binary prediction sets are limited; fuzzy prediction sets offer richer guarantees.
method Generalize prediction sets to fuzzy sets, showing they are e-values with merging properties.
result Optimal e-values lead to optimal fuzzy prediction sets, including optimal conformal prediction.
This paper re-examines conformal e-prediction and its advantages over conformal prediction.
problem The relationship between conformal prediction and conformal e-prediction.
method Systematic re-examination of conformal prediction and conformal e-prediction from a modern perspective.
result Conformal e-prediction has advantages such as ease of designing conditional predictors and guaranteed validity of cross-predictors.
Two new methods improve efficiency of conformal predictive systems.
problem Efficiency of conformal predictive systems in regression problems.
method Split conformal predictive systems and cross-conformal predictive systems.
result Cross-conformal predictive systems are more efficient but not guaranteed valid.
Self-calibrating conformal prediction improves interval efficiency and offers a practical alternative.
problem Improving the reliability and uncertainty quantification of machine learning predictions.
method Combines Venn-Abers calibration and conformal prediction for binary and regression problems.
result Improves interval efficiency through model calibration and offers practical alternatives.
Study uses deep learning to predict asset prices, finds complex target processes lead to meaningless predictions.
problem Complexity of successful price prediction models hinders understanding.
method Deep learning models for high-frequency price prediction, focusing on volatility and directional prediction.
result Inadequately defined target price process renders predictions meaningless.
The paper emphasizes the importance of joint predictions over marginal predictions for decision-making.
problem The need for accurate joint predictions in decision-making problems.
method The paper analyzes combinatorial decision problems, sequential predictions, and multi-armed bandits, introducing an approximate Thompson sampling algorithm and new regret bounds.
result Accurate joint predictions are essential for good performance in decision-making problems.
Behavior modification improves prediction accuracy by nudging user behavior.
problem Improving prediction accuracy using behavior modification techniques.
method Combining prediction and behavior modification with reinforcement learning algorithms.
result Behavior modification can make predictions more certain but may not generalize.
Predictions can shape outcomes, study helps predict these effects.
problem Understanding how predictions influence real-world outcomes.
method Causal identifiability analysis of prediction-covariate-outcome relationships.
result Standard supervised learning can identify transferable relationships from predictions.
Prediction problems often admit competing models that perform almost equally well. This effect challenges key assumptions in machine learning when competing models assign conflicting predictions. In this paper, we define predictive multiplicity as the ability of a prediction problem to admit competing models with confl…
Proposes feature conformal prediction for broader application in semantic feature spaces.
problem Establishing valid prediction intervals in semantic feature spaces.
method Extends conformal prediction to semantic feature spaces using deep representation learning.
result Feature conformal prediction outperforms regular conformal prediction under mild assumptions.
Acute kidney injury (AKI) commonly occurs in hospitalized patients and can lead to serious medical complications. In order to optimally predict AKI before it develops at any time during a hospital stay, we present a novel framework in which AKI is continually predicted automatically from EHR data over the entire hospit…
AutoCP automates the construction of accurate prediction intervals.
problem Creating valid and accurate prediction intervals for machine learning models.
method AutoML framework that optimizes prediction interval length for better accuracy and less conservatism.
result AutoCP significantly outperforms benchmark algorithms in constructing accurate prediction intervals.
Proposes a method to apply conformal prediction to probabilistic time series forecasting models.
problem Obtaining accurate prediction regions for multi-step time series forecasting with probabilistic models.
method Conformalises conditional normalising flows to generate potentially disjoint prediction regions.
result Improves predictive efficiency in time series forecasting with multimodal distributions.
ICP improves prediction intervals for continuous outcomes at lower computational cost.
problem Systematic bias in point predictions that undermines their use in decision-making.
method Develops Isotonic Conformal Prediction (ICP) framework to decouple calibration from prediction-set construction.
result SICP and TICP procedures match SC-CP coverage at lower computational cost.
Optimizes predictions for specific tasks using parametrized decision analysis.
problem Optimizing predictions for specific decision tasks of interest.
method Designs a class of parametrized actions for Bayesian decision analysis.
result Derives efficient and interpretable solutions for various action parametrizations and loss functions.
New measures for prediction validity and consonant plausibility introduced.
problem Challenges in predicting future observations and quantifying prediction uncertainty.
method Introducing Type-2 validity and using consonant plausibility measures and conformal prediction.
result Achieving both Type-1 and Type-2 validity through consonant plausibility measures and conformal prediction.
RFpredInterval package builds prediction intervals for random forests and boosted forests.
problem Quantifying uncertainty in random forest and boosted forest point predictions.
method 16 methods to build prediction intervals with random forests and boosted forests.
result The proposed method outperforms existing methods in building prediction intervals.
FPPI selectively uses predictions to improve inference efficiency.
problem Improving statistical inference with limited labeled data and heterogeneous prediction quality.
method Filtered Prediction-Powered Inference (FPPI) framework.
result FPPI achieves strictly improved asymptotic efficiency compared to existing methods.
COP improves online conformal prediction by incorporating data patterns, leading to tighter prediction sets.
problem Overly conservative prediction sets in online conformal prediction methods when data distribution shifts.
method Conformal Optimistic Prediction (COP) incorporating estimated cumulative distribution function of non-conformity scores.
result COP produces tighter prediction sets with valid coverage guarantees, outperforming other methods.
Unified framework for estimating random forest prediction errors.
problem Estimating prediction errors for random forests.
method Novel estimator of conditional prediction error distribution function.
result Proposed estimators enable competitive prediction intervals.
Most existing examples of full conformal predictive systems, split-conformal predictive systems, and cross-conformal predictive systems impose severe restrictions on the adaptation of predictive distributions to the test object at hand. In this paper we develop split-conformal and cross-conformal predictive systems tha…
This paper studies trade-offs in private prediction methods.
problem Leakage of training data information in machine learning predictions.
method Private training and private prediction methods with trade-offs.
result Private training methods outperform private prediction methods in various settings.
Conformal prediction helps quantify uncertainty but its use by humans is unclear.
problem Uncertainty quantification in predictions for human decision making.
method Decision theoretic framework for evaluating predictive uncertainty.
result Conformal prediction sets and human decision making goals are in tension.
New tree splitting criteria improve probabilistic predictions.
problem Improving tree-based nonparametric predictive distributions.
method Using proper scoring rules for tree splitting criteria.
result Trees with new splitting criteria produce better predictive distributions.