Infinite attention: NNGP and NTK for deep attention networks

by   Jiri Hron, et al.

There is a growing amount of literature on the relationship between wide neural networks (NNs) and Gaussian processes (GPs), identifying an equivalence between the two for a variety of NN architectures. This equivalence enables, for instance, accurate approximation of the behaviour of wide Bayesian NNs without MCMC or variational approximations, or characterisation of the distribution of randomly initialised wide NNs optimised by gradient descent without ever running an optimiser. We provide a rigorous extension of these results to NNs involving attention layers, showing that unlike single-head attention, which induces non-Gaussian behaviour, multi-head attention architectures behave as GPs as the number of heads tends to infinity. We further discuss the effects of positional encodings and layer normalisation, and propose modifications of the attention mechanism which lead to improved results for both finite and infinitely wide NNs. We evaluate attention kernels empirically, leading to a moderate improvement upon the previous state-of-the-art on CIFAR-10 for GPs without trainable kernels and advanced data preprocessing. Finally, we introduce new features to the Neural Tangents library (Novak et al., 2020) allowing applications of NNGP/NTK models, with and without attention, to variable-length sequences, with an example on the IMDb reviews dataset.


Bayesian Convolutional Neural Networks with Many Channels are Gaussian Processes

There is a previously identified equivalence between wide fully connecte...

Interpretable deep Gaussian processes

We propose interpretable deep Gaussian Processes (GPs) that combine the ...

State Space representation of non-stationary Gaussian Processes

The state space (SS) representation of Gaussian processes (GP) has recen...

On the expected behaviour of noise regularised deep neural networks as Gaussian processes

Recent work has established the equivalence between deep neural networks...

An Overview of Uncertainty Quantification Methods for Infinite Neural Networks

To better understand the theoretical behavior of large neural networks, ...

On Exact Computation with an Infinitely Wide Neural Net

How well does a classic deep net architecture like AlexNet or VGG19 clas...

Geometric Path Enumeration for Equivalence Verification of Neural Networks

As neural networks (NNs) are increasingly introduced into safety-critica...