An unfeasability view of neural network learning

01/04/2022

∙

We define the notion of a continuously differentiable perfect learning algorithm for multilayer neural network architectures and show that such algorithms don't exist provided that the length of the data set exceeds the number of involved parameters and the activation functions are logistic, tanh or sin.

READ FULL TEXT

An unfeasability view of neural network learning

An Analysis of State-of-the-art Activation Functions For Supervised Deep Neural Network

Otimizacao de pesos e funcoes de ativacao de redes neurais aplicadas na previsao de series temporais

DeepLABNet: End-to-end Learning of Deep Radial Basis Networks with Fully Learnable Basis Functions

A comparative study of back propagation and its alternatives on multilayer perceptrons

Evolution of Activation Functions: An Empirical Investigation

Neural network feedback controller for inertial platform

Architopes: An Architecture Modification for Composite Pattern Learning, Increased Expressiveness, and Reduced Training Time

An unfeasability view of neural network learning

Related Research

An Analysis of State-of-the-art Activation Functions For Supervised Deep Neural Network

Otimizacao de pesos e funcoes de ativacao de redes neurais aplicadas na previsao de series temporais

DeepLABNet: End-to-end Learning of Deep Radial Basis Networks with Fully Learnable Basis Functions

A comparative study of back propagation and its alternatives on multilayer perceptrons

Evolution of Activation Functions: An Empirical Investigation

Neural network feedback controller for inertial platform

Architopes: An Architecture Modification for Composite Pattern Learning, Increased Expressiveness, and Reduced Training Time