arXiv research

New algorithm tackles multi-armed bandit with arbitrary delays and general bounded losses.

New method reduces regret for sparse adversarial SSP problems.

LEGW improves large-batch training for both CNNs and RNNs.

We consider the problem of provably optimal exploration in reinforcement learning for finite horizon MDPs. We show that an optimistic modification to value iteration achieves a regret bound of $\tilde{O}( \sqrt{HSAT} + H^2S^2A+H\sqrt{T})$ where $H$ is the time horizon, $S$ the number of states, $A$ the number of action…

We introduce a new weight-decay scaling rule to maintain sublayer gains across different widths in modern scale-invariant architectures.

Study finds the number of modes in Gaussian kernel density estimators scales with sqrt(β log β).

New algorithm bounds MAB regret for unknown scale and magnitude of losses.

ResNets converge to a limit model with improved error rates.

Study optimizes KSD estimation from samples, revealing Hilbert-Schmidt vs trace scales.

New algorithm reduces robust optimization scale for better constraint satisfaction.

Improved bound for Gaussian mechanism in differential privacy.

We design differentially private algorithms for the problem of online linear optimization in the full information and bandit settings with optimal $\tilde{O}(\sqrt{T})$ regret bounds. In the full-information setting, our results demonstrate that $ε$-differential privacy may be ensured for free -- in particular, the reg…

Online and stochastic learning has emerged as powerful tool in large scale optimization. In this work, we generalize the Douglas-Rachford splitting (DRs) method for minimizing composite functions to online and stochastic settings (to our best knowledge this is the first time DRs been generalized to sequential version).…

New method improves likelihood-free parameter estimation in complex models.

Improved online algorithm for convex losses with near-optimal swap regret.

Stochastic variational inference (SVI) employs stochastic optimization to scale up Bayesian computation to massive data. Since SVI is at its core a stochastic gradient-based algorithm, horizontal parallelism can be harnessed to allow larger scale inference. We propose a lock-free parallel implementation for SVI which a…

We analyze the eigenvalue distribution of a neural network's kernel under specific scaling.

A new measure of model complexity based on Fisher Information.

Upper and lower bounds on regret for noisy optimization of Brownian motion.

New method reduces ensemble size for linear bandits, achieving near optimal regret.

Improved cumulative regret for graph-based stochastic linear bandits.

Paper derives a fast learning rate for deep neural networks without scale invariant activation functions.

Improved kernel Stein discrepancy for large-scale data.

New algorithm handles MDPs with unknown, changing rewards efficiently.

A new law limits kurtosis contrast in balanced mixtures.

The paper examines rigidity of thin domains under specific boundary conditions.

New algorithm reduces submodular maximization regret to nearly optimal bounds.

We study the generalization properties of ridge regression with random features in the statistical learning framework. We show for the first time that $O(1/\sqrt{n})$ learning bounds can be achieved with only $O(\sqrt{n}\log n)$ random features rather than $O({n})$ as suggested by previous results. Further, we prove fa…

We propose a new random pruning method (called "submodular sparsification (SS)") to reduce the cost of submodular maximization. The pruning is applied via a "submodularity graph" over the $n$ ground elements, where each directed edge is associated with a pairwise dependency defined by the submodular function. In each s…

New algorithm tackles batched stochastic linear bandits with 1-bit communication constraints.

We construct a new map from a convex function to a distribution on its domain, with the property that this distribution is a multi-scale exploration of the function. We use this map to solve a decade-old open problem in adversarial bandit convex optimization by showing that the minimax regret for this problem is $\tild…

The study compares uniform-price and discriminatory auctions in terms of learning difficulty.

New algorithm reduces regret from sqrt(T) to polylog(T) in stochastic contextual linear bandits.

Optimizing over-the-air convex optimization, analog schemes are nearly optimal at low SNR.

The paper establishes theoretical foundations for low-rank knowledge distillation in LLMs.

New algorithm reduces regret in linear bandits by nearly optimal factors.

We derive an online learning algorithm with improved regret guarantees for `easy' loss sequences. We consider two types of `easiness': (a) stochastic loss sequences and (b) adversarial loss sequences with small effective range of the losses. While a number of algorithms have been proposed for exploiting small effective…

We study the problem of adaptive control of a high dimensional linear quadratic (LQ) system. Previous work established the asymptotic convergence to an optimal controller for various adaptive control schemes. More recently, for the average cost LQ problem, a regret bound of ${O}(\sqrt{T})$ was shown, apart form logarit…

Kernel methods provide a principled way to perform non linear, nonparametric learning. They rely on solid functional analytic foundations and enjoy optimal statistical properties. However, at least in their basic form, they have limited applicability in large scale scenarios because of stringent computational requireme…

Motivated by applications of large-scale graph clustering, we study random-walk-based LOCAL algorithms whose running times depend only on the size of the output cluster, rather than the entire graph. All previously known such algorithms guarantee an output conductance of $\tilde{O}(\sqrt{φ(A)})$ when the target set $A$…

New model for personalized online advertising with multi-user interaction.

This paper extends combinatorial semi-bandits to graph feedback, improving regret bounds.

We consider the problem of online combinatorial optimization under semi-bandit feedback, where a learner has to repeatedly pick actions from a combinatorial decision set in order to minimize the total losses associated with its decisions. After making each decision, the learner observes the losses associated with its a…

Classifies low-energy harmonic maps from curved surfaces to spheres.

Filtering data with a pre-trained model improves multimodal contrastive learning performance.

Variant of mSSA improves time series prediction error.

SpectralTS improves efficiency of Thompson Sampling for graph-based bandits.

TSAW improves MCMC integral estimation with faster convergence.