arXiv research

Extends InfoGAIL to improve imitation learning over extended periods.

Optimal algorithms for non-linear ridge bandits reduce burn-in cost.

Study fast learning rates for square loss in dependent data with hypercontractivity condition.

New RL algorithms reduce costs for single-agent and federated learning.

New algorithm learns optimal policies with minimal memory and time.

Parallelized bandit algorithms speed up decision-making.

New algorithm reduces cold-start costs in multi-armed bandits for many products.

SGD shows distinct phases in learning single-index models, achieving optimal sample complexity and regret.

ULA estimates covariance of log-concave distributions efficiently.

Observer learns optimal policy from learner's actions without rewards.

This paper is concerned with offline reinforcement learning (RL), which learns using pre-collected data without further exploration. Effective offline RL would be able to accommodate distribution shift and limited data coverage. However, prior algorithms or analyses either suffer from suboptimal sample complexities or …

A new algorithm reduces memory and computational needs for reinforcement learning.

New bounds on trajectory safety in training models with Langevin Dynamics.

GD with large, adaptive stepsizes achieves optimal risk in logistic regression.

Communication costs, resulting from synchronization requirements during learning, can greatly slow down many parallel machine learning algorithms. In this paper, we present a parallel Markov chain Monte Carlo (MCMC) algorithm in which subsets of data are processed independently, with very little communication. First, w…

Two algorithms learn Gaussian graphical models from Glauber dynamics trajectories.

We present a general framework for accelerating a large class of widely used Markov chain Monte Carlo (MCMC) algorithms. Our approach exploits fast, iterative approximations to the target density to speculatively evaluate many potential future steps of the chain in parallel. The approach can accelerate computation of t…

Optimizes MCMC output compression by selecting a subset of states.

Study bounds noise level in linear regression with dependent data.

In this paper we address the following question: Can we approximately sample from a Bayesian posterior distribution if we are only allowed to touch a small mini-batch of data-items for every sample we generate?. An algorithm based on the Langevin equation with stochastic gradients (SGLD) was previously proposed to solv…

Restricted Boltzmann machines (RBMs) are powerful machine learning models, but learning and some kinds of inference in the model require sampling-based approximations, which, in classical digital computers, are implemented using expensive MCMC. Physical computation offers the opportunity to reduce the cost of sampling …

Curriculum learning improves model training efficiency and performance.

Enhanced MH algorithm reduces expensive function evaluations and improves sampling efficiency.

UCRL3 improves UCRL2's efficiency in reinforcement learning by reducing exploration.

Particle Metropolis-Hastings (PMH) allows for Bayesian parameter inference in nonlinear state space models by combining Markov chain Monte Carlo (MCMC) and particle filtering. The latter is used to estimate the intractable likelihood. In its original formulation, PMH makes use of a marginal MCMC proposal for the parame…

Particle MCMC is a class of algorithms that can be used to analyse state-space models. They use MCMC moves to update the parameters of the models, and particle filters to propose values for the path of the state-space model. Currently the default is to use random walk Metropolis to update the parameter values. We show …

New algorithm finds critical points in non-convex optimization with heavy-tailed gradients.

New GLB algorithm handles non-stationary data with forgetting.

This work analyzes the convergence rate of unrolling for optimizing quadratic objectives.

This study provides an explicit expansion of KL divergence's gradient flow in Fisher-Rao geometry.

Regularized Stein thinning improves MCMC output approximations.

Bayesian realized EGARCH models improve tail risk forecasting.

New analysis shows SGD with noise doesn't leak more privacy with more iterations.

New compression methods handle biased input sequences for more accurate posterior summaries.

The paper analyzes Adam and SGD in nonstationary optimization, revealing tradeoffs between noise and drift.

The paper improves importance sampling and MCMC methods for complex distributions.

Spectral gradient methods outperform Euclidean in certain deep learning scenarios.

Near-optimal tests and confidence sequences for non-parametric data.

New RL algorithm reduces sample complexity for optimal learning.

BART improves predictive performance but slows down with more data.

New algorithm reduces regret in infinite MDPs with optimal variance-dependent bounds.

D-REX learns reward functions from ranked demonstrations to beat the demonstrator's performance.

LEARN-SAM improves RL from sub-optimal demonstrations by localizing expert policies and selectively using demonstrations.

Unified reinforcement learning from imperfect demonstrations improves robustness.

New method improves IL from imperfect demos using confidence scores.

A framework helps reinforcement learning agents understand and decompose tasks from human demonstrations.

ARLD uses active deep Q-learning to learn faster with expert demonstrations.

Novel framework learns policies from noisy expert demonstrations.