Skip to article frontmatterSkip to article content

Feed-Forward Network

Definition

A Feed-Forward Network (FFN) is a fundamental neural network architecture where information flows in one direction, from input to output layers, through one or more hidden layers, without any cycles or loops. Each neuron in a layer is connected to all neurons in the subsequent layer, but there are no connections between neurons in the same layer or backwards connections. FFNs are also known as Multi-Layer Perceptrons (MLPs) and form basic building blocks in more complex architectures like transformers.

Tags

Neural Networks, Architecture, Basic Concepts, MLP

References

  • Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533-536. Rumelhart et al. (1986)
References
  1. Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533–536. 10.1038/323533a0