General Purpose
Online Neural Networks

Online Neural Network

Neural Network implementation using Hedge Backpropagation based on Online Deep Learning: Learning Deep Neural Networks on the Fly1.

Parameters

  • max_num_hidden_layers(Default 10) → The maximum number of hidden layers
  • qtd_neuron_hidden_layer(Default: 32) → Hidden dimension of the intermediate neural network layers.
  • n_classes(int) → Number of classes.
  • b(Default: 0.99) → Discounting parameter in the hedge backprop algorithm.
  • n(Default: 0.01) → Learning rate parameter in the hedge backprop algorithm.
  • s(Default: 0.2) → Smoothing parameter in the hedge backprop algorithm.

Example Usage

We can create an instance and deploy ONN model like this.

import turboml as tb
model = tb.ONN(n_classes=2)

Footnotes

  1. D. Sahoo, Q. Pham, J. Lu and S. Hoi. Online Deep Learning: Learning Deep Neural Networks on the Fly (opens in a new tab) ↩

Neural NetworksONNX