Learning viewpoint planning in active recognition on a small sampling budget: a Kriging approach

Abstract : This paper focuses on viewpoint planning for 3D active object recognition. The objective is to design a planning policy into a Q-learning framework with a limited number of samples. Most existing stochastic techniques are therefore inapplicable. We propose to use Kriging and Bayesian Optimization coupled with Q-learning to obtain a computationally-efficient viewpoint-planning design, under a restrictive sampling budget. Experimental results on a representative database, including a comparison with classical approaches, show promising results for this strategy.
Complete list of metadatas

Cited literature [15 references]  Display  Hide  Download

https://hal-supelec.archives-ouvertes.fr/hal-00520814
Contributor : Julien Marzat <>
Submitted on : Thursday, September 8, 2011 - 10:09:35 AM
Last modification on : Tuesday, March 26, 2019 - 2:28:03 PM
Long-term archiving on : Tuesday, November 13, 2012 - 10:06:19 AM

File

DTIM1041.1291040181.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00520814, version 1

Citation

Joseph Defretin, Julien Marzat, Hélène Piet-Lahanier. Learning viewpoint planning in active recognition on a small sampling budget: a Kriging approach. 9th IEEE Conference on Machine Learning and Applications, ICMLA 2010, Dec 2010, Washington, D.C., United States. pp.169-174. ⟨hal-00520814⟩

Share

Metrics

Record views

619

Files downloads

324