Abstract | ||
---|---|---|
Gaussian processes (GPs) enable a probabilistic approach to important estimation and classification tasks that arise in robotics applications. Meanwhile, most GP-based methods are often prohibitively slow, thereby posing a substantial barrier to practical applications. Existing "sparse" methods to speed up GPs seek to either make the model more sparse, or find ways to more efficiently manage a large covariance matrix. In this paper, we present an orthogonal approach that memoises (i.e. reuses) previous computations in GP inference. We demonstrate that a substantial speedup can be achieved by incorporating memoisation into applications in which GPs must be updated frequently. Moreover, we derive a novel online update scheme for sparse GPs that can be used in conjunction with our memoisation approach for a synergistic improvement in performance. Across three robotic vision applications, we demonstrate between 40-100% speed-up over the standard method for inference in GP mixtures. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1109/ICRA40945.2020.9196734 | 2020 IEEE International Conference on Robotics and Automation (ICRA) |
Keywords | DocType | Volume |
GP mixtures,data association,Gaussian processes,probabilistic approach,important estimation,classification tasks,robotics applications,GP-based methods,sparse methods,covariance matrix,orthogonal approach,GP inference,online update scheme,sparse GPs,memoisation approach,robotic vision applications | Conference | 2020 |
Issue | ISSN | ISBN |
1 | 1050-4729 | 978-1-7281-7396-2 |
Citations | PageRank | References |
0 | 0.34 | 5 |
Authors | ||
5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ki Myung Brian Lee | 1 | 1 | 2.37 |
Wolfram Martens | 2 | 10 | 1.23 |
Jayant Khatkar | 3 | 0 | 0.34 |
Robert Fitch | 4 | 323 | 38.97 |
Ramgopal R. Mettu | 5 | 177 | 22.23 |