Sparse Kernel Gaussian Processes through Iterative Charted Refinement (ICR)
This addresses the computational bottleneck for researchers and practitioners using GPs in large-scale applications, representing a novel method rather than an incremental improvement.
The paper tackles the computational complexity of Gaussian Processes (GPs) by introducing Iterative Charted Refinement (ICR), a generative method that achieves O(N) time for decaying kernels without nested optimizations, outperforming existing methods by one order of magnitude in speed and scaling to 122 billion parameters.
Gaussian Processes (GPs) are highly expressive, probabilistic models. A major limitation is their computational complexity. Naively, exact GP inference requires $\mathcal{O}(N^3)$ computations with $N$ denoting the number of modeled points. Current approaches to overcome this limitation either rely on sparse, structured or stochastic representations of data or kernel respectively and usually involve nested optimizations to evaluate a GP. We present a new, generative method named Iterative Charted Refinement (ICR) to model GPs on nearly arbitrarily spaced points in $\mathcal{O}(N)$ time for decaying kernels without nested optimizations. ICR represents long- as well as short-range correlations by combining views of the modeled locations at varying resolutions with a user-provided coordinate chart. In our experiment with points whose spacings vary over two orders of magnitude, ICR's accuracy is comparable to state-of-the-art GP methods. ICR outperforms existing methods in terms of computational speed by one order of magnitude on the CPU and GPU and has already been successfully applied to model a GP with $122$ billion parameters.