%0 Generic %1 yang2019tensor %A Yang, Greg %D 2019 %K 2019 arxiv deep-learning machine-learning %T Tensor Programs I: Wide Feedforward or Recurrent Neural Networks of Any Architecture are Gaussian Processes %U http://arxiv.org/abs/1910.12478 %X Wide neural networks with random weights and biases are Gaussian processes, as originally observed by Neal (1995) and more recently by Lee et al. (2018) and Matthews et al. (2018) for deep fully-connected networks, as well as by Novak et al. (2019) and Garriga-Alonso et al. (2019) for deep convolutional networks. We show that this Neural Network-Gaussian Process correspondence surprisingly extends to all modern feedforward or recurrent neural networks composed of multilayer perceptron, RNNs (e.g. LSTMs, GRUs), (nD or graph) convolution, pooling, skip connection, attention, batch normalization, and/or layer normalization. More generally, we introduce a language for expressing neural network computations, and our result encompasses all such expressible neural networks. This work serves as a tutorial on the *tensor programs* technique formulated in Yang (2019) and elucidates the Gaussian Process results obtained there. We provide open-source implementations of the Gaussian Process kernels of simple RNN, GRU, transformer, and batchnorm+ReLU network at github.com/thegregyang/GP4A.

@misc{yang2019tensor, abstract = {Wide neural networks with random weights and biases are Gaussian processes, as originally observed by Neal (1995) and more recently by Lee et al. (2018) and Matthews et al. (2018) for deep fully-connected networks, as well as by Novak et al. (2019) and Garriga-Alonso et al. (2019) for deep convolutional networks. We show that this Neural Network-Gaussian Process correspondence surprisingly extends to all modern feedforward or recurrent neural networks composed of multilayer perceptron, RNNs (e.g. LSTMs, GRUs), (nD or graph) convolution, pooling, skip connection, attention, batch normalization, and/or layer normalization. More generally, we introduce a language for expressing neural network computations, and our result encompasses all such expressible neural networks. This work serves as a tutorial on the *tensor programs* technique formulated in Yang (2019) and elucidates the Gaussian Process results obtained there. We provide open-source implementations of the Gaussian Process kernels of simple RNN, GRU, transformer, and batchnorm+ReLU network at github.com/thegregyang/GP4A.}, added-at = {2019-12-11T15:11:40.000+0100}, author = {Yang, Greg}, biburl = {https://www.bibsonomy.org/bibtex/284730a6b824d67410e8150c44c8db11a/analyst}, description = {[1910.12478] Tensor Programs I: Wide Feedforward or Recurrent Neural Networks of Any Architecture are Gaussian Processes}, interhash = {1a02b2d848a36bb6dbc6943ff5fe3773}, intrahash = {84730a6b824d67410e8150c44c8db11a}, keywords = {2019 arxiv deep-learning machine-learning}, note = {cite arxiv:1910.12478Comment: Appearing in NeurIPS 2019; 10 pages of main text; 12 figures, 11 programs; 73 pages total}, timestamp = {2019-12-11T15:11:40.000+0100}, title = {Tensor Programs I: Wide Feedforward or Recurrent Neural Networks of Any Architecture are Gaussian Processes}, url = {http://arxiv.org/abs/1910.12478}, year = 2019 }