Machine Learning

PSPS: a novel task-agnostic framework enables valid and efficient ML-assisted statistical inference for virtually any task, simply using summary statistics from existing analysis routines!

Researchers unveil the Infeasibility Theorem, proving optimal class-incremental learning is impossible with discriminative models due to task confusion, and the Feasibility Theorem, showing generative…

This paper proposes novel algorithms achieving near-optimal regret in adversarial and logarithmic regret in stochastic multi-armed bandit settings with heavy-tailed losses, relaxing strong assumptions…

FedPLVM tames cross-domain variance in federated prototype learning using dual-level clustering and an a-sparsity loss, achieving superior performance.

Reinforcement learning achieves unprecedented fast convergence rates in continuous state-action spaces by leveraging novel stability properties of Markov Decision Processes.

Shortcut back-propagation and an evolutionary training framework conquer gradient vanishing in spiking neural networks, drastically improving training and achieving state-of-the-art accuracy.

TabEBM: Class-specific EBMs boost tabular data augmentation, improving classification accuracy, especially on small datasets, by generating high-quality synthetic data.

Leveraging symmetry in automated equation discovery improves accuracy and simplicity of learned governing equations, enhancing robustness against noise and achieving higher success rates across divers…

Researchers unveil a novel algorithm for high-dimensional symmetric linear bandits, achieving a regret bound of O(d^(2/3)T^(2/3)log(d)), surpassing limitations of existing approaches that assume expli…

Swift Sampler (SS) automates the learning of efficient data samplers for deep learning, achieving significant performance gains (e.g., 1.5% on ImageNet) with minimal computational cost using only 10 p…

Deep learning’s Adam-style optimizers exhibit a surprising surge phenomenon: optimal learning rates initially increase, then decrease, before converging to a non-zero value as batch size grows.

SLATE: Supra-Laplacian encoding for spatio-temporal Transformers achieves state-of-the-art dynamic link prediction by innovatively using a multi-layer graph representation and a unique cross-attention…

Supervised Kernel Thinning accelerates kernel regression by cleverly compressing data, achieving quadratic speedups in training and inference with minimal accuracy loss.

Neural network hyperparameter transferability across vastly different model sizes is achieved via a newly discovered property called ‘Super Consistency’ of loss landscapes.

This paper introduces Subwords as Skills (SaS), a fast and efficient skill extraction method for sparse-reward reinforcement learning that uses tokenization. SaS enables 1000x faster skill extraction…

SubgDiff enhances molecular representation learning by incorporating substructural information into a diffusion model framework, achieving superior performance in molecular force predictions.

Sub-optimal expert data improves Inverse Reinforcement Learning by significantly reducing ambiguity in reward function estimation.

SAUE: A novel multi-source blended-target domain adaptation approach using style adaptation and uncertainty estimation to improve model robustness and accuracy.

Sumba: a novel forecasting model achieves up to 8.5% improvement by using a structured matrix basis to generate dynamic spatial structures with lower variance and better interpretability.

SICSM, a novel framework, integrates selective SSMs and GFNs to accurately infer complex dynamical system structures from irregularly sampled, partially observed trajectories.