Výsledky vyhledávání

Report

A Quantitative Approach to Predicting Representational Learning and Performance in Neural Networks

Autor: Pyle, Ryan, Musslick, Sebastian, Cohen, Jonathan D., Patel, Ankit B.

A key property of neural networks (both biological and artificial) is how they learn to represent and manipulate input information in order to solve a task. Different types of representations may be suited to different types of tasks, making identify

Externí odkaz: http://arxiv.org/abs/2307.07575

Zobrazit plný text záznamu

Report

Shallow Univariate ReLu Networks as Splines: Initialization, Loss Surface, Hessian, & Gradient Flow Dynamics

Autor: Sahs, Justin, Pyle, Ryan, Damaraju, Aneel, Caro, Josue Ortega, Tavaslioglu, Onur, Lu, Andy, Patel, Ankit

Understanding the learning dynamics and inductive bias of neural networks (NNs) is hindered by the opacity of the relationship between NN parameters and the function represented. We propose reparametrizing ReLU NNs as continuous piecewise linear spli

Externí odkaz: http://arxiv.org/abs/2008.01772

Zobrazit plný text záznamu

Report

Local Convolutions Cause an Implicit Bias towards High Frequency Adversarial Examples

Autor: Caro, Josue Ortega, Ju, Yilong, Pyle, Ryan, Dey, Sourav, Brendel, Wieland, Anselmi, Fabio, Patel, Ankit

Adversarial Attacks are still a significant challenge for neural networks. Recent work has shown that adversarial perturbations typically contain high-frequency features, but the root cause of this phenomenon remains unknown. Inspired by theoretical

Externí odkaz: http://arxiv.org/abs/2006.11440

Zobrazit plný text záznamu

Report

A model of reward-modulated motor learning with parallelcortical and basal ganglia pathways

Autor: Pyle, Ryan, Rosenbaum, Robert

Many recent studies of the motor system are divided into two distinct approaches: Those that investigate how motor responses are encoded in cortical neurons' firing rate dynamics and those that study the learning rules by which mammals and songbirds

Externí odkaz: http://arxiv.org/abs/1803.03304

Zobrazit plný text záznamu

Akademický článek

Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.

Report

Spatiotemporal dynamics and reliable computations in recurrent spiking neural networks

Autor: Pyle, Ryan, Rosenbaum, Robert

Publikováno v: Phys. Rev. Lett. 118, 018103 (2017)

Randomly connected networks of excitatory and inhibitory spiking neurons provide a parsimonious model of neural variability, but are notoriously unreliable for performing computations. We show that this difficulty is overcome by incorporating the wel

Externí odkaz: http://arxiv.org/abs/1611.01557

Zobrazit plný text záznamu

Report

Highly connected neurons spike less frequently in balanced networks

Autor: Pyle, Ryan, Rosenbaum, Robert

Publikováno v: Phys. Rev. E 93, 040302 (2016)

Many biological neuronal networks exhibit highly variable spiking activity. Balanced networks offer a parsimonious model of this variability. In balanced networks, strong excitatory synaptic inputs are canceled by strong inhibitory inputs on average

Externí odkaz: http://arxiv.org/abs/1601.04972

Zobrazit plný text záznamu