Policy Search with High-Dimensional Context Variables
This work addresses a bottleneck in applying contextual policy search to real-world tasks with high-dimensional inputs, offering an incremental improvement over existing methods.
The paper tackles the problem of learning policies from high-dimensional context variables like camera images, where naive dimensionality reduction fails to retain task-relevant information. The proposed method integrates supervised linear dimensionality reduction with nuclear norm regularization into a model-based policy search framework, outperforming PCA and a state-of-the-art baseline in experiments.
Direct contextual policy search methods learn to improve policy parameters and simultaneously generalize these parameters to different context or task variables. However, learning from high-dimensional context variables, such as camera images, is still a prominent problem in many real-world tasks. A naive application of unsupervised dimensionality reduction methods to the context variables, such as principal component analysis, is insufficient as task-relevant input may be ignored. In this paper, we propose a contextual policy search method in the model-based relative entropy stochastic search framework with integrated dimensionality reduction. We learn a model of the reward that is locally quadratic in both the policy parameters and the context variables. Furthermore, we perform supervised linear dimensionality reduction on the context variables by nuclear norm regularization. The experimental results show that the proposed method outperforms naive dimensionality reduction via principal component analysis and a state-of-the-art contextual policy search method.