Neural networks algorithms take inspiration from the brain and involve large numers of simple neurons all working together. They are typically organised in layers where the outputs of one layer of neurons form the input to the next, but some neural networks, including Boltzmann machine and Kohonan networks, have connections within layers. Many neural networks are trained using some form of backpropagation.
Used in Chap. 1: pages 3, 6; Chap. 2: page 20; Chap. 3: page 35; Chap. 5: page 69; Chap. 7: pages 100, 101, 105, 107; Chap. 8: pages 111, 115, 124; Chap. 9: pages 127, 128, 130, 132, 134, 137; Chap. 10: pages 151, 152; Chap. 11: pages 159, 160, 169, 170, 172; Chap. 12: pages 173, 192, 195, 196, 201, 202; Chap. 13: pages 217, 218; Chap. 14: pages 221, 229, 231, 233, 236, 237; Chap. 15: page 253; Chap. 16: pages 261, 266; Chap. 18: pages 292, 300, 302, 307; Chap. 19: pages 313, 323, 324, 329; Chap. 20: pages 343, 344; Chap. 21: pages 351, 356, 357, 360; Chap. 22: pages 365, 368, 369, 371; Chap. 23: pages 384, 389
Also known as neural net
A multi-layer perceptron architecture
Deep learning architecture – multiple layers, with varying connection topologies.
Boltzmann Machine – fully conneced between and withi layers