Výsledky vyhledávání

Report

Weak Convergence Analysis of Online Neural Actor-Critic Algorithms

Autor: Lam, Samuel Chun-Hei, Sirignano, Justin, Wang, Ziheng

We prove that a single-layer neural network trained with the online actor critic algorithm converges in distribution to a random ordinary differential equation (ODE) as the number of hidden units and the number of training steps $\rightarrow \infty$.

Externí odkaz: http://arxiv.org/abs/2403.16825

Zobrazit plný text záznamu

Report

Deep Neural Network Initialization with Sparsity Inducing Activations

Autor: Price, Ilan, Ball, Nicholas Daultry, Lam, Samuel C. H., Jones, Adam C., Tanner, Jared

Inducing and leveraging sparse activations during training and inference is a promising avenue for improving the computational efficiency of deep networks, which is increasingly important as network sizes continue to grow and their application become

Externí odkaz: http://arxiv.org/abs/2402.16184

Zobrazit plný text záznamu

Report

Kernel Limit of Recurrent Neural Networks Trained on Ergodic Data Sequences

Autor: Lam, Samuel Chun-Hei, Sirignano, Justin, Spiliopoulos, Konstantinos

Mathematical methods are developed to characterize the asymptotics of recurrent neural networks (RNN) as the number of hidden units, data samples in the sequence, hidden state updates, and training steps simultaneously grow to infinity. In the case o

Externí odkaz: http://arxiv.org/abs/2308.14555

Zobrazit plný text záznamu