Machine Learning

Random noise pretraining dramatically speeds up and enhances neural network learning without weight transport, mimicking the brain’s developmental process and achieving performance comparable to backp…

CoDA-NO, a novel neural operator, revolutionizes multiphysics PDE solving via codomain tokenization, enabling efficient self-supervised pretraining and few-shot learning for superior generalization.

Pretrained Optimization Model (POM) excels at zero-shot black-box optimization, outperforming existing methods, especially in high dimensions, through direct application or few-shot fine-tuning.

Eliciting high-dimensional probability distributions from experts using only preference comparisons is achieved via normalizing flows and a novel functional prior, resolving the problem of collapsing …

PreTS algorithm efficiently identifies the most preferred policy in bandit problems with vector-valued rewards, achieving asymptotically optimal sample complexity.

Human-like choice modeling revolutionizes preference learning! A new tractable model, CRCS, significantly improves utility inference from human data, outperforming existing methods.

Preference Flow Matching (PFM) streamlines preference integration into pre-trained models using flow matching, overcoming fine-tuning limitations and enabling robust alignment with human preferences.

Predictive Attractor Models (PAM) offer a biologically-plausible, streaming sequence memory architecture that avoids catastrophic forgetting and generates multiple future possibilities.

Boosting label distribution learning accuracy and efficiency, this research proposes using ternary labels instead of binary labels to predict label distributions, thus enhancing annotation efficiency …

Deep learning algorithms now predict quantum ground state properties with constant sample complexity, regardless of system size, improving upon previous methods.

New analysis reveals how reweighted least-squares algorithms for linear diagonal networks achieve favorable performance in high-dimensional settings, improving upon existing theoretical guarantees and…

Researchers achieved high-accuracy differentially private (DP) models by using a novel DP continual pre-training strategy with only 10% public data, mitigating the performance degradation common in DP…

MGPO: Efficient online RL adaptation via prompt optimization of pre-trained multi-goal transformers.

Revolutionizing unordered data compression, this paper introduces autoregressive shuffle coding, achieving state-of-the-art speeds and compression rates on massive datasets.

PS-BAX, a novel Bayesian algorithm execution method using posterior sampling, efficiently selects evaluation points for complex tasks, outperforming existing methods in speed and scalability.

Post-hoc model transformations can reverse performance trends, prompting a reevaluation of model selection strategies and suggesting a new ‘post-hoc selection’ method for improved model development.

Policy-Shaped Prediction (PSP) improves model-based reinforcement learning by focusing world models on task-relevant information, significantly enhancing robustness against distracting stimuli.

First-order policy optimization for robust average-cost MDPs achieves linear convergence with increasing step size and 0(1/ε) complexity with constant step size, solving a critical gap in existing res…

Boosting reinforcement learning, this paper introduces h-PMD, a novel algorithm enhancing policy mirror descent with lookahead for faster convergence and improved sample complexity.

Pin-Tuning: A parameter-efficient method for few-shot molecular property prediction that significantly improves accuracy with fewer trainable parameters via in-context tuning and Bayesian weight cons…