empirical risk minimization
PulseAugur coverage of empirical risk minimization — every cluster mentioning empirical risk minimization across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
ReliableNet tackles confident-wrong predictions in deep learning
Researchers have introduced ReliableNet, a novel approach to trustworthy classification in deep learning. This method directly constrains the probability of a prediction being both confident and incorrect, a critical fa…
-
New research explores robust PAC learning under Cressie--Read divergences
Researchers have published a paper detailing the sample complexity of distributionally robust PAC learning, specifically focusing on Cressie--Read divergences. The study establishes new bounds for hypothesis classes wit…
-
New framework tackles spurious correlations in deep learning models · 2 sources tracked
Researchers have developed a novel two-stage framework to improve the robustness of deep neural networks against distribution shifts by addressing spurious correlations. The method first uses generative intervention wit…
-
New framework uses Gibbs measures for data-driven hierarchical learning
Researchers have developed a novel data-driven framework for learning systems that utilizes Gibbs measures on hierarchical structures. This approach transforms the empirical loss function into an interaction potential, …
-
Two papers analyze theoretical limits of empirical risk minimization in ML
Two new research papers explore the theoretical underpinnings of empirical risk minimization (ERM) in machine learning. The first paper, "Replica Symmetry Breaking and Algorithmic Thresholds in Empirical Risk Minimizati…
-
New Differentially Private Algorithm for Weighted Empirical Risk Minimization Developed
Researchers have developed a new differentially private algorithm for weighted empirical risk minimization (wERM), a generalization of standard ERM that accounts for varying individual contributions to the objective fun…
-
New research tackles domain generalization challenges in Human Activity Recognition
A new research paper explores the challenges of domain generalization in Human Activity Recognition (HAR) due to distribution shifts. The study systematically evaluates four types of shifts—device type, sensor placement…
-
New privacy framework 'predictability' complements differential privacy
Researchers have introduced a new privacy framework called "privacy via predictability" that offers a more fine-grained approach than traditional differential privacy (DP). This new method accounts for an attacker's spe…
-
New Research Tackles Privacy-Preserving Ad Conversion Prediction
A new research paper on arXiv introduces a method for statistical learning from attribution sets, addressing privacy constraints in advertising domains where direct links between ad clicks and conversions are unavailabl…
-
Research paper unifies CoCoA and ADMM optimization algorithms
A new research paper explores the relationship between two families of distributed optimization algorithms, CoCoA and ADMM. By unifying them through a primal-dual perspective, the study reveals that certain ADMM variant…
-
New Research Unveils Fundamental Limits of k-Fold Cross-Validation
A new research paper explores the theoretical limitations of k-fold cross-validation, a widely used technique for estimating the performance of machine learning models. The study, focusing on the majority algorithm in b…
-
New bound links generalization gap to data entropy
Researchers have developed a new method to bound the generalization gap in machine learning models, which is a key factor in understanding overfitting. This novel approach establishes a model-independent upper bound for…
-
New paper proposes multi-axis fairness for toxicity detection models
A new paper introduces a framework for evaluating fairness in toxicity detection models, considering ranking, calibration, and abstention. The research found that standard training methods like Empirical Risk Minimizati…
-
New framework improves U-statistics with active inference for costly labels
Researchers have developed a new active inference framework for U-statistics, aiming to improve estimation efficiency when labeling data is expensive. This approach selectively queries informative labels within a fixed …
-
New Quadratic Objective Perturbation method enhances differential privacy for ML
Researchers have introduced Quadratic Objective Perturbation (QOP) as a novel method for differential privacy in machine learning. Unlike Linear Objective Perturbation (LOP), which requires bounded gradients, QOP uses a…
-
AI researchers develop PAC-learning algorithm for consensus elicitation
Researchers have developed a new theoretical framework called Probably Approximately Consensus to identify broadly agreeable ideas on online platforms. This approach models consensus as an interval within a one-dimensio…