arXiv research

A single policy suffices for near-optimal parallel exploration in RL.

Parallelized bandit algorithms speed up decision-making.

We consider parallel asynchronous Markov Chain Monte Carlo (MCMC) sampling for problems where we can leverage (stochastic) gradients to define continuous dynamics which explore the target distribution. We outline a solution strategy for this setting based on stochastic gradient Hamiltonian Monte Carlo sampling (SGHMC) …

Unified framework for randomized exploration in cooperative MARL.

Can one parallelize complex exploration exploitation tradeoffs? As an example, consider the problem of optimal high-throughput experimental design, where we wish to sequentially design batches of experiments in order to simultaneously learn a surrogate function mapping stimulus to response and identify the maximum of t…

The paper explores parallel 1-forms on special Finsler manifolds and their properties.

A scalable portfolio approach speeds up Bayesian optimization for noisy functions.

Monte Carlo (MC) methods are widely used for Bayesian inference and optimization in statistics, signal processing and machine learning. A well-known class of MC methods are Markov Chain Monte Carlo (MCMC) algorithms. In order to foster better exploration of the state space, specially in high-dimensional applications, s…

In this paper, we consider the challenge of maximizing an unknown function f for which evaluations are noisy and are acquired with high cost. An iterative procedure uses the previous measures to actively select the next estimation of f which is predicted to be the most useful. We focus on the case where the function ca…

Parallel sentences are a relatively scarce but extremely useful resource for many applications including cross-lingual retrieval and statistical machine translation. This research explores our methodology for mining such data from previously obtained comparable corpora. The task is highly practical since non-parallel m…

Batch Thompson Sampling reduces exploration-exploitation trade-off in online decision making.

Investigates nearly Kähler and parallel G2 manifolds using Hitchin functionals.

Study explores geometric structure and prior for beta-logistic distribution.

The paper explores Lorentzian connections with parallel skew torsion.

Monte Carlo Tree Search (MCTS) algorithms have achieved great success on many challenging benchmarks (e.g., Computer Go). However, they generally require a large number of rollouts, making their applications costly. Furthermore, it is also extremely challenging to parallelize MCTS due to its inherent sequential nature:…

Chemical space is so large that brute force searches for new interesting molecules are infeasible. High-throughput virtual screening via computer cluster simulations can speed up the discovery process by collecting very large amounts of data in parallel, e.g., up to hundreds or thousands of parallel measurements. Bayes…

Parallel sentences are a relatively scarce but extremely useful resource for many applications including cross-lingual retrieval and statistical machine translation. This research explores our new methodologies for mining such data from previously obtained comparable corpora. The task is highly practical since non-para…

This research optimizes energy consumption forecasting in Puno using parallel computing and ARIMA models.

TensorOpt finds optimal parallelization strategies for DNN training.

Study perturbations of submodules in Drury-Arveson space, finding smooth vector bundles with Hermitian connections.

TSSM splits neural networks for parallel training with minimal accuracy loss.

Ensembled neural networks improve MRI image quality.

PLN-Nets with two linear layers and parallel LN achieve universal approximation.

In real world industrial applications of topic modeling, the ability to capture gigantic conceptual space by learning an ultra-high dimensional topical representation, i.e., the so-called "big model", is becoming the next desideratum after enthusiasms on "big data", especially for fine-grained downstream tasks such as …

When training large machine learning models with many variables or parameters, a single machine is often inadequate since the model may be too large to fit in memory, while training can take a long time even with stochastic updates. A natural recourse is to turn to distributed cluster computing, in order to harness add…

Deploying deep learning (DL) models across multiple compute devices to train large and complex models continues to grow in importance because of the demand for faster and more frequent training. Data parallelism (DP) is the most widely used parallelization strategy, but as the number of devices in data parallel trainin…

There is significant recent interest to parallelize deep learning algorithms in order to handle the enormous growth in data and model sizes. While most advances focus on model parallelization and engaging multiple computing agents via using a central parameter server, aspect of data parallelization along with decentral…

Study uses supercomputers to improve financial predictions.

New submersions found in nearly Kähler geometry.

Safe RL for autonomous vehicles using PCPO with trust regions and parallel learners.

The paper explores $K^α$-translators on parallel and canal surfaces in 3D space.

Inspired by the recent work of Physicists Hertog-Horowitz-Maeda, we prove two stability results for compact Riemannian manifolds with nonzero parallel spinors. Our first result says that Ricci flat metrics which also admits nonzero parallel spinors are stable (in the direction of changes in conformal structures) as the…

We study the question of whether parallelization in the exploration of the feasible set can be used to speed up convex optimization, in the local oracle model of computation. We show that the answer is negative for both deterministic and randomized algorithms applied to essentially any of the interesting geometries and…

We study a type of connection forms, given by Chen integrals, over pathspaces by placing such forms within a category-theoretic framework of principal bundles and connections. We introduce a notion of 'decorated' principal bundles, develop parallel transport on such bundles, and explore specific examples in the context…

SOBER optimizes and quadrates efficiently in parallel for diverse tasks.

New algorithm tackles resource allocation in multi-armed bandits to balance speed and throughput.

The paper explores biconservative surfaces in a 4D sphere, finding a unique family of non-isometric surfaces.

New method solves blind inverse problems by optimizing both operator and image parameters.

The paper explores almost paracomplex structures on 4-manifolds and their properties.

Study of time-dependent metrics and connections in geometry.

New approach for reward-free exploration reduces estimation error.

ICSGLD improves efficiency in posterior sampling for big data.

Sideways trains video models by overwriting activations as new frames arrive, potentially improving generalization.

This paper explores parallels between minimal surfaces and Einstein manifolds.

Unified framework for statistical inference in gradient boosting regression.

Lecture notes on reinforcement learning using statistical methods.

The Dirichlet process (DP) is a fundamental mathematical tool for Bayesian nonparametric modeling, and is widely used in tasks such as density estimation, natural language processing, and time series modeling. Although MCMC inference methods for the DP often provide a gold standard in terms asymptotic accuracy, they ca…

For G_2-manifolds the Fernández-Gray class X_1+X_4 is shown to consist of the union of the class X_4 of G_2-manifolds locally conformal to parallel G_2-structures and that of conformal transformations of nearly parallel or weak holonomy G_2-manifolds of type X_1. The analogous conclusion is obtained for Gray-Hervella c…