US 11,673,560 B2
Efficient computational inference using Gaussian processes
Vincent Dutordoir, Cambridge (GB); James Hensman, Cambridge (GB); and Nicolas Durrande, Cambridge (GB)
Assigned to SECONDMIND LIMITED, Cambridge (GB)
Filed by SECONDMIND LIMITED, Cambridgeshire (GB)
Filed on Jun. 3, 2021, as Appl. No. 17/338,158.
Application 17/338,158 is a continuation of application No. 16/836,116, filed on Mar. 31, 2020, granted, now 11,027,743.
Prior Publication US 2021/0300390 A1, Sep. 30, 2021
Int. Cl. B60W 40/12 (2012.01); G05B 13/02 (2006.01); B60W 50/04 (2006.01); B60W 10/30 (2006.01); B60W 10/06 (2006.01); B60W 10/18 (2012.01)
CPC B60W 40/12 (2013.01) [B60W 50/04 (2013.01); G05B 13/028 (2013.01); B60W 10/06 (2013.01); B60W 10/18 (2013.01); B60W 10/30 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system comprising:
a simulation module configured to generate a simulation of a physical system having a plurality of adjustable parameters; and
an analysis module configured to:
determine a plurality of candidate sets of values for the adjustable parameters; and
for each of the candidate sets of values for the plurality of adjustable parameters:
obtain, from the generated simulation of the physical system, predictions of one or more characteristics of the physical system;
determine a performance prediction for the physical system based on the obtained predictions of the one or more characteristics of the physical system; and
generate a data point having an input portion indicative of the candidate set of values and an output portion indicative of the determined performance prediction, the input portion having a first number of dimensions;
augment each data point to include an additional dimension comprising a bias value;
project each augmented data point onto a surface of a unit hypersphere of the first number of dimensions;
determine, using the projected augmented data points, a set of parameter values for a sparse variational Gaussian process, GP, on said unit hypersphere, the sparse variational GP having a zonal kernel and depending on a set of inducing variables randomly distributed according to a multi-dimensional Gaussian distribution and each corresponding to a reproducing kernel Hilbert space inner product between the GP and a spherical harmonic of the first number of dimensions; and
determine, using the sparse variational GP with the determined set of parameter values, a further set of values for the plurality of adjustable parameters of the physical system.