arXiv subjects

Clayton Webster

Publications and source records attributed to Clayton Webster.

At least 19 recordsLinked to original sources

arXiv2024

Anisotropic Gaussian Smoothing for Gradient-based Optimization

This article introduces a novel family of optimization algorithms - Anisotropic Gaussian Smoothing Gradient Descent (AGS-GD), AGS-Stochastic Gradient Descent (AGS-SGD), and AGS-Adam - that employ anisotropic Gaussian smoothing to enhance traditional gradient-based methods, including GD, SGD, and Adam. The primary goal of these approaches is to address the challenge of optimization methods becoming trapped in suboptimal local minima by replacing the standard gradient with a non-local gradient derived from averaging function values using anisotropic Gaussian smoothing. Unlike isotropic Gaussian smoothing (IGS), AGS adapts the smoothing directionality based on the properties of the underlying function, aligning better with complex loss landscapes and improving convergence. The anisotropy is computed by adjusting the covariance matrix of the Gaussian distribution, allowing for directional smoothing tailored to the gradient's behavior. This technique mitigates the impact of minor fluctuations, enabling the algorithms to approach global minima more effectively. We provide detailed convergence analyses that extend the results from both the original (unsmoothed) methods and the IGS case to the more general anisotropic smoothing, applicable to both convex and non-convex, L-smooth functions. In the stochastic setting, these algorithms converge to a noisy ball, with its size determined by the smoothing parameters. The article also outlines the theoretical benefits of anisotropic smoothing and details its practical implementation using Monte Carlo estimation, aligning with established zero-order optimization techniques.

math.OC↗

arXiv2024

Mamba for Scalable and Efficient Personalized Recommendations

In this effort, we propose using the Mamba for handling tabular data in personalized recommendation systems. We present the \textit{FT-Mamba} (Feature Tokenizer\,$+$\,Mamba), a novel hybrid model that replaces Transformer layers with Mamba layers within the FT-Transformer architecture, for handling tabular data in personalized recommendation systems. The \textit{Mamba model} offers an efficient alternative to Transformers, reducing computational complexity from quadratic to linear by enhancing the capabilities of State Space Models (SSMs). FT-Mamba is designed to improve the scalability and efficiency of recommendation systems while maintaining performance. We evaluate FT-Mamba in comparison to a traditional Transformer-based model within a Two-Tower architecture on three datasets: Spotify music recommendation, H\&M fashion recommendation, and vaccine messaging recommendation. Each model is trained on 160,000 user-action pairs, and performance is measured using precision (P), recall (R), Mean Reciprocal Rank (MRR), and Hit Ratio (HR) at several truncation values. Our results demonstrate that FT-Mamba outperforms the Transformer-based model in terms of computational efficiency while maintaining or exceeding performance across key recommendation metrics. By leveraging Mamba layers, FT-Mamba provides a scalable and effective solution for large-scale personalized recommendation systems, showcasing the potential of the Mamba architecture to enhance both efficiency and accuracy.

cs.IR↗

arXiv2023

Gaussian smoothing gradient descent for minimizing functions (GSmoothGD)

This work analyzes the convergence of a class of smoothing-based gradient descent methods when applied to optimization problems. In particular, Gaussian smoothing is employed to define a nonlocal gradient that reduces high-frequency noise, small variations, and rapid fluctuations in the computation of the descent directions while preserving the structure and features of the loss landscape. The resulting Gaussian smoothing gradient descent (GSmoothGD) approach can facilitate gradient descent in navigating away from and avoiding local minima with increased ease, thereby substantially enhancing its overall performance even when applied to non-convex optimization problems. This work also provides rigorous theoretical error estimates on the rate of convergence of GSmoothGD iterates. These estimates exemplify the impact of underlying function convexity, smoothness, input dimension, and the Gaussian smoothing radius. To combat the curse of dimensionality, we numerically approximate the GSmoothGD nonlocal gradient using Monte Carlo (MC) sampling and provide a theory in which the iterates converge regardless of the function smoothness and dimension. Finally, we present several strategies to update the smoothing parameter aimed at diminishing the impact of local minima, thereby rendering the attainment of global minima more achievable. Computational evidence complements the present theory and shows the effectiveness of the MC-GSmoothGD method compared to other smoothing-based algorithms, momentum-based approaches, and classical gradient-based algorithms from numerical optimization.

math.OC↗

arXiv2023

Improved Performance of Stochastic Gradients with Gaussian Smoothing

This paper formalizes and analyzes Gaussian smoothing applied to two prominent optimization methods: Stochastic Gradient Descent (GSmoothSGD) and Adam (GSmoothAdam) in deep learning. By attenuating small fluctuations, Gaussian smoothing lowers the risk of gradient-based algorithms converging to poor local minima. These methods simplify the loss landscape while boosting robustness to noise and improving generalization, helping base algorithms converge more effectively to global minima. Existing approaches often rely on zero-order approximations, which increase training time due to inefficiencies in automatic differentiation. To address this, we derive Gaussian-smoothed loss functions for feedforward and convolutional networks, improving computational efficiency. Numerical experiments demonstrate the enhanced performance of our smoothing algorithms over unsmoothed counterparts, confirming the theoretical benefits.

math.OC↗

arXiv2023

Increasing Entropy to Boost Policy Gradient Performance on Personalization Tasks

In this effort, we consider the impact of regularization on the diversity of actions taken by policies generated from reinforcement learning agents trained using a policy gradient. Policy gradient agents are prone to entropy collapse, which means certain actions are seldomly, if ever, selected. We augment the optimization objective function for the policy with terms constructed from various $\varphi$-divergences and Maximum Mean Discrepancy which encourages current policies to follow different state visitation and/or action choice distribution than previously computed policies. We provide numerical experiments using MNIST, CIFAR10, and Spotify datasets. The results demonstrate the advantage of diversity-promoting policy regularization and that its use on gradient-based approaches have significantly improved performance on a variety of personalization tasks. Furthermore, numerical evidence is given to show that policy regularization increases performance without losing accuracy.

cs.LG↗

arXiv2021

On the Unreasonable Efficiency of State Space Clustering in Personalization Tasks

In this effort we consider a reinforcement learning (RL) technique for solving personalization tasks with complex reward signals. In particular, our approach is based on state space clustering with the use of a simplistic $k$-means algorithm as well as conventional choices of the network architectures and optimization algorithms. Numerical examples demonstrate the efficiency of different RL procedures and are used to illustrate that this technique accelerates the agent's ability to learn and does not restrict the agent's performance.

cs.LG↗

arXiv2021

On the strong convergence of forward-backward splitting in reconstructing jointly sparse signals

We consider the problem of reconstructing an infinite set of sparse, finite-dimensional vectors, that share a common sparsity pattern, from incomplete measurements. This is in contrast to the work [17], where the single vector signal can be infinite-dimensional, and [28], which extends the aforementioned work to the joint sparse recovery of finite number of infinite-dimensional vectors. In our case, to take account of the joint sparsity and promote the coupling of nonvanishing components, we employ a convex relaxation approach with mixed norm penalty $\ell_{2,1}$. This paper discusses the computation of the solutions of linear inverse problems with such relaxation by a forward-backward splitting algorithm. However, since the solution matrix possesses infinitely many columns, the arguments of [17] no longer apply. As such, we establish new strong convergence results for the algorithm, in particular when the set of jointly sparse vectors is infinite.

math.OC↗

arXiv2021

Greedy Shallow Networks: An Approach for Constructing and Training Neural Networks

We present a greedy-based approach to construct an efficient single hidden layer neural network with the ReLU activation that approximates a target function. In our approach we obtain a shallow network by utilizing a greedy algorithm with the prescribed dictionary provided by the available training data and a set of possible inner weights. To facilitate the greedy selection process we employ an integral representation of the network, based on the ridgelet transform, that significantly reduces the cardinality of the dictionary and hence promotes feasibility of the greedy selection. Our approach allows for the construction of efficient architectures which can be treated either as improved initializations to be used in place of random-based alternatives, or as fully-trained networks in certain cases, thus potentially nullifying the need for backpropagation training. Numerical experiments demonstrate the tenability of the proposed concept and its advantages compared to the conventional techniques for selecting architectures and initializations for neural networks.

cs.LG↗

arXiv2021

An asymptotically compatible probabilistic collocation method for randomly heterogeneous nonlocal problems

In this paper we present an asymptotically compatible meshfree method for solving nonlocal equations with random coefficients, describing diffusion in heterogeneous media. In particular, the random diffusivity coefficient is described by a finite-dimensional random variable or a truncated combination of random variables with the Karhunen-Loève decomposition, then a probabilistic collocation method (PCM) with sparse grids is employed to sample the stochastic process. On each sample, the deterministic nonlocal diffusion problem is discretized with an optimization-based meshfree quadrature rule. We present rigorous analysis for the proposed scheme and demonstrate convergence for a number of benchmark problems, showing that it sustains the asymptotic compatibility spatially and achieves an algebraic or sub-exponential convergence rate in the random coefficients space as the number of collocation points grows. Finally, to validate the applicability of this approach we consider a randomly heterogeneous nonlocal problem with a given spatial correlation structure, demonstrating that the proposed PCM approach achieves substantial speed-up compared to conventional Monte Carlo simulations.

math.NA↗

arXiv2021

Offline Policy Comparison under Limited Historical Agent-Environment Interactions

We address the challenge of policy evaluation in real-world applications of reinforcement learning systems where the available historical data is limited due to ethical, practical, or security considerations. This constrained distribution of data samples often leads to biased policy evaluation estimates. To remedy this, we propose that instead of policy evaluation, one should perform policy comparison, i.e. to rank the policies of interest in terms of their value based on available historical data. In addition we present the Limited Data Estimator (LDE) as a simple method for evaluating and comparing policies from a small number of interactions with the environment. According to our theoretical analysis, the LDE is shown to be statistically reliable on policy comparison tasks under mild assumptions on the distribution of the historical data. Additionally, our numerical experiments compare the LDE to other policy evaluation methods on the task of policy ranking and demonstrate its advantage in various settings.

cs.LG↗

arXiv2021

Robust learning with implicit residual networks

In this effort, we propose a new deep architecture utilizing residual blocks inspired by implicit discretization schemes. As opposed to the standard feed-forward networks, the outputs of the proposed implicit residual blocks are defined as the fixed points of the appropriately chosen nonlinear transformations. We show that this choice leads to the improved stability of both forward and backward propagations, has a favorable impact on the generalization power and allows to control the robustness of the network with only a few hyperparameters. In addition, the proposed reformulation of ResNet does not introduce new parameters and can potentially lead to a reduction in the number of required layers due to improved forward stability. Finally, we derive the memory-efficient training algorithm, propose a stochastic regularization technique and provide numerical results in support of our findings.

cs.LG↗

arXiv2020

Neural network integral representations with the ReLU activation function

In this effort, we derive a formula for the integral representation of a shallow neural network with the ReLU activation function. We assume that the outer weighs admit a finite $L_1$-norm with respect to Lebesgue measure on the sphere. For univariate target functions we further provide a closed-form formula for all possible representations. Additionally, in this case our formula allows one to explicitly solve the least $L_1$-norm neural network representation for a given function.

cs.LG↗

arXiv2019

Analysis of Deep Neural Networks with Quasi-optimal polynomial approximation rates

We show the existence of a deep neural network capable of approximating a wide class of high-dimensional approximations. The construction of the proposed neural network is based on a quasi-optimal polynomial approximation. We show that this network achieves an error rate that is sub-exponential in the number of polynomial functions, $M$, used in the polynomial approximation. The complexity of the network which achieves this sub-exponential rate is shown to be algebraic in $M$.

math.NA↗

arXiv2019

The Natural Greedy Algorithm for reduced bases in Banach spaces

In this effort we introduce and analyze a novel reduced basis approach, used to construct an approximating subspace for a given set of data. Our technique, which we call the Natural Greedy Algorithm (NGA), is based on a recursive approach for iteratively constructing such subspaces, and coincides with the standard, and the extensively studied, Orthogonal Greedy Algorithm (OGA) in a Hilbert space. However, for a given set of data in a general Banach space, the NGA is straightforward to implement and overcomes the explosion in computational effort introduced by the OGA, as we utilize an entirely new technique for projecting onto appropriate subspaces. We provide a rigorous analysis of the NGA, and demonstrate that it's theoretical performance is similar to the OGA, while the realization of the former results in significant computational savings through a substantially improved numerical procedure. Furthermore, we show that the empirical interpolation method (EIM) can be viewed as a special case of the NGA. Finally, several numerical examples are used to illustrate the advantages of our NGA compared with other greedy algorithms and additional popular reduced bases methods, including EIM and proper orthogonal decomposition.

math.FA↗

arXiv2019

Analytic Continuation of Noisy Data Using Adams Bashforth ResNet

We propose a data-driven learning framework for the analytic continuation problem in numerical quantum many-body physics. Designing an accurate and efficient framework for the analytic continuation of imaginary time using computational data is a grand challenge that has hindered meaningful links with experimental data. The standard Maximum Entropy (MaxEnt)-based method is limited by the quality of the computational data and the availability of prior information. Also, the MaxEnt is not able to solve the inversion problem under high level of noise in the data. Here we introduce a novel learning model for the analytic continuation problem using a Adams-Bashforth residual neural network (AB-ResNet). The advantage of this deep learning network is that it is model independent and, therefore, does not require prior information concerning the quantity of interest given by the spectral function. More importantly, the ResNet-based model achieves higher accuracy than MaxEnt for data with higher level of noise. Finally, numerical examples show that the developed AB-ResNet is able to recover the spectral function with accuracy comparable to MaxEnt where the noise level is relatively small.

physics.comp-ph↗

arXiv2019

Reconstructing high-dimensional Hilbert-valued functions via compressed sensing

We present and analyze a novel sparse polynomial technique for approximating high-dimensional Hilbert-valued functions, with application to parameterized partial differential equations (PDEs) with deterministic and stochastic inputs. Our theoretical framework treats the function approximation problem as a joint sparse recovery problem, where the set of jointly sparse vectors is possibly infinite. To achieve the simultaneous reconstruction of Hilbert-valued functions in both parametric domain and Hilbert space, we propose a novel mixed-norm based $\ell_1$ regularization method that exploits both energy and sparsity. Our approach requires extensions of concepts such as the restricted isometry and null space properties, allowing us to prove recovery guarantees for sparse Hilbert-valued function reconstruction. We complement the enclosed theory with an algorithm for Hilbert-valued recovery, based on standard forward-backward algorithm, meanwhile establishing its strong convergence in the considered infinite-dimensional setting. Finally, we demonstrate the minimal sample complexity requirements of our approach, relative to other popular methods, with numerical experiments approximating the solutions of high-dimensional parameterized elliptic PDEs.

math.NA↗

arXiv2019

A class of null space conditions for sparse recovery via nonconvex, non-separable minimizations

For the problem of sparse recovery, it is widely accepted that nonconvex minimizations are better than $\ell_1$ penalty in enhancing the sparsity of solution. However, to date, the theory verifying that nonconvex penalties outperform (or are at least as good as) $\ell_1$ minimization in exact, uniform recovery has mostly been limited to separable cases. In this paper, we establish general recovery guarantees through null space conditions for nonconvex, non-separable regularizations, which are slightly less demanding than the standard null space property for $\ell_1$ minimization.

math.OC↗

arXiv2018

A mixed $\ell_1$ regularization approach for sparse simultaneous approximation of parameterized PDEs

We present and analyze a novel sparse polynomial technique for the simultaneous approximation of parameterized partial differential equations (PDEs) with deterministic and stochastic inputs. Our approach treats the numerical solution as a jointly sparse reconstruction problem through the reformulation of the standard basis pursuit denoising, where the set of jointly sparse vectors is infinite. To achieve global reconstruction of sparse solutions to parameterized elliptic PDEs over both physical and parametric domains, we combine the standard measurement scheme developed for compressed sensing in the context of bounded orthonormal systems with a novel mixed-norm based $\ell_1$ regularization method that exploits both energy and sparsity. In addition, we are able to prove that, with minimal sample complexity, error estimates comparable to the best $s$-term and quasi-optimal approximations are achievable, while requiring only a priori bounds on polynomial truncation error with respect to the energy norm. Finally, we perform extensive numerical experiments on several high-dimensional parameterized elliptic PDE models to demonstrate the superior recovery properties of the proposed approach.

math.NA↗