arXiv research

Derives time-averaged active inference from control principles.

Study the averaging principle for non-autonomous slow-fast systems and apply it to financial local stochastic volatility models.

New Q-learning method achieves optimal sample complexity for average-reward problems.

Study optimizes step size for Metropolis algorithm in non-identifiable cases.

SNAP improves robust computation by emphasizing trustworthy items and downweighting outliers.

EC method calibrates neural networks by matching average confidence to correct label proportion.

New algorithms improve estimation of treatment effects.

In this paper, we prove that the L^2 Betti numbers of an amenable covering space can be approximated by the average Betti numbers of a regular exhaustion, under some hypotheses. We also prove that some L^2 spectral invariants can be approximated by the corresponding average spectral invariants of a regular exhaustion. …

The paper rethinks the use of exponential averaging in machine learning optimization.

The paper analyzes MACD using operator theory.

This paper addresses the problem of prediction with expert advice for outcomes in a geodesic space with non-positive curvature in the sense of Alexandrov. Via geometric considerations, and in particular the notion of barycenters, we extend to this setting the definition and analysis of the classical exponentially weigh…

This dissertation advances the theoretical foundation of local optimization methods in Federated Learning.

FP-BMA improves generalization by encouraging flat posteriors in Bayesian Model Averaging.

We define On-Average KL-Privacy and present its properties and connections to differential privacy, generalization and information-theoretic quantities including max-information and mutual information. The new definition significantly weakens differential privacy, while preserving its minimalistic design features such …

Revises Bayesian model averaging for foundation models.

New algorithms for learning MDPs with linear approximations in infinite-horizon settings.

This paper develops several average-case reduction techniques to show new hardness results for three central high-dimensional statistics problems, implying a statistical-computational gap induced by robustness, a detection-recovery gap and a universality principle for these gaps. A main feature of our approach is to ma…

Consider vector valued harmonic maps of at most linear growth, defined on a complete non-compact Riemannian manifold with non-negative Ricci curvature. For the norm square of the pull-back of the target volume form by such maps, we report a strong maximum principle, and equalities among its supremum, its asymptotic ave…

Study optimal offline RL with uncertainty sets and distribution shifts.

A new uncertainty principle helps traders better understand market activity.

Study non-rectangular robust MDPs for average-reward, finding optimal policies and transient values.

This article revisits an analysis on inaccuracies of time series averaging under dynamic time warping conducted by \cite{Niennattrakul2007}. The authors presented a correctness-criterion and introduced drift-outs of averages from clusters. They claimed that averages are inaccurate if they are incorrect or drift-outs. F…

Consider a family of portfolio strategies with the aim of achieving the asymptotic growth rate of the best one. The idea behind Cover's universal portfolio is to build a wealth-weighted average which can be viewed as a buy-and-hold portfolio of portfolios. When an optimal portfolio exists, the wealth-weighted average c…

We present in this paper a new premium computation principle based on the use of prior information from multiple sources for computing the premium charged to a policyholder. Under this framework, based on the use of Ordered Weighted Averaging (OWA) operators, we propose alternative collective and Bayes premiums and des…

This work improves Q-learning for average-reward MDPs, reducing sample and communication complexities in federated settings.

Unordered feature sets are a nonstandard data structure that traditional neural networks are incapable of addressing in a principled manner. Providing a concatenation of features in an arbitrary order may lead to the learning of spurious patterns or biases that do not actually exist. Another complication is introduced …

In this paper we study the probabilistic properties of the posteriors in a speech recognition system that uses a deep neural network (DNN) for acoustic modeling. We do this by reducing Kaldi's DNN shared pdf-id posteriors to phone likelihoods, and using test set forced alignments to evaluate these using a calibration s…

New perspective on federated learning as posterior inference, improving optimization.

The paper introduces moment multicalibration for estimating uncertainty across subgroups.

Develops a dynamical method to prove the sharp Berezin-Li-Yau inequality.

New model-free RL algorithm tackles robust average-reward problems with finite sample complexity analysis.

The Kapustin-Witten equations on R^4 are equations for a pair of connection on the product principle SU(2) bundle and 1-form with values in the product Lie algebra bundle. The 1-form is the Higgs field. A dichotomy is proved to the effect that either the averaged norm of the Higgs field on large radius spheres grows fa…

A new method identifies causal direction using dense functional classes.

FedAc accelerates Federated Averaging for distributed optimization.

New method SF-AdamW trains large models without decay phases or memory overhead.

A new framework evaluates large language models efficiently and accurately.

Study large deviations in life insurance portfolios without identical distributions.

We consider in this work a system of two stochastic differential equations named the perturbed compositional gradient flow. By introducing a separation of fast and slow scales of the two equations, we show that the limit of the slow motion is given by an averaged ordinary differential equation. We then demonstrate that…

A method for calculating multi-portfolio time consistent multivariate risk measures in discrete time is presented. Market models for $d$ assets with transaction costs or illiquidity and possible trading constraints are considered on a finite probability space. The set of capital requirements at each time and state is c…

Combining different models is a widely used paradigm in machine learning applications. While the most common approach is to form an ensemble of models and average their individual predictions, this approach is often rendered infeasible by given resource constraints in terms of memory and computation, which grow linearl…

The paper establishes principles for initializing and designing GNNs with ReLU activations to avoid oversmoothing and correlation collapse.

The motivation for this paper is to apply Bayesian structure learning using Model Averaging in large-scale networks. Currently, Bayesian model averaging algorithm is applicable to networks with only tens of variables, restrained by its super-exponential complexity. We present a novel framework, called LSBN(Large-Scale …

The paper shows averaging gradients leads to memorization, proposing an alternative algorithm to focus on invariances.

Motivated by the scarcity of accurate payoff feedback in practical applications of game theory, we examine a class of learning dynamics where players adjust their choices based on past payoff observations that are subject to noise and random disturbances. First, in the single-player case (corresponding to an agent tryi…

Meta-learning model predicts intervention effects from uncertain causal graphs.

We justify and give error estimates for binomial approximations of game (Israeli) options in the Black--Scholes market with Lipschitz continuous path dependent payoffs which are new also for usual American style options. We show also that rational (optimal) exercise times and hedging self-financing portfolios of binomi…

BayesIMP combines multiple causal graphs to estimate average treatment effects with uncertainty.

The paper connects convex functions to p-subharmonic functions and proves their equivalence.