deep neural network

Terms from Artificial Intelligence: humans at the heart of algorithms

A deep neural network is a neural network with lots of layers.. Tyoically the layers are also large, in the sense of lots of nodes. Each layer can also have a differnt size and use different types of learning. For example, it is common for the first layer to be a restricted Boltzmann machine in order to perform dimensional reduction. Typically the inner laters are underdetermined (many equally good arrangements of weightings); this and theor disrance from the output layer means that backpropagation or similar learning rules need to have very slow learning rates to avoid instabilities. In addition, more layers and laregr layers usually requires more training data. Together these mean that deep learning (training of deep neural networks) requires. a lot of training data. This combination of computational cost and data volume is one of the main reasons that the use deep neural networks were not widely adopted for many years.

Used in Chap. 1: page 7; Chap. 6: pages 76, 80, 81, 83, 84; Chap. 7: page 98; Chap. 8: pages 102, 103, 105, 106, 115, 116; Chap. 9: page 129; Chap. 11: pages 148, 157, 158, 160; Chap. 12: pages 182, 183; Chap. 13: pages 201, 203; Chap. 14: pages 212, 217; Chap. 17: page 265; Chap. 18: pages 283, 289; Chap. 19: pages 301, 310; Chap. 20: pages 315, 317; Chap. 21: pages 330, 331, 333, 334, 336, 339, 341; Chap. 22: pages 343, 346; Chap. 23: pages 369, 370; Chap. 24: page 375

Also known as deep learning

Deep learning architecture – multiple layers, with varying connection topologies