Look where you look! Saliency-guided Q-networks for generalization in visual Reinforcement Learning
About
Deep reinforcement learning policies, despite their outstanding efficiency in simulated visual control tasks, have shown disappointing ability to generalize across disturbances in the input training images. Changes in image statistics or distracting background elements are pitfalls that prevent generalization and real-world applicability of such control policies. We elaborate on the intuition that a good visual policy should be able to identify which pixels are important for its decision, and preserve this identification of important sources of information across images. This implies that training of a policy with small generalization gap should focus on such important pixels and ignore the others. This leads to the introduction of saliency-guided Q-networks (SGQN), a generic method for visual reinforcement learning, that is compatible with any value function learning method. SGQN vastly improves the generalization capability of Soft Actor-Critic agents and outperforms existing stateof-the-art methods on the Deepmind Control Generalization benchmark, setting a new reference in terms of training efficiency, generalization gap, and policy interpretability.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Continuous Control | DMC-GB video hard | Cartpole Swingup Score5.44e+4 | 18 | |
| Continuous Control | DMC-GB video easy | Cartpole Swingup Score717 | 12 | |
| Finger Spin | DMControl Novel view (test) | Reward553.3 | 12 | |
| Cup Catch | DMControl Novel view (test) | Reward803 | 12 | |
| Robotic manipulation (Reach) | Robotic-Manipulation reach (test2) | Performance33 | 7 | |
| Cheetah Run | DMControl-GB color-easy (test) | Average Episode Return312 | 7 | |
| Robotic Manipulation | peg-in-box (test2) | Return194 | 7 | |
| Robotic Manipulation | peg-in-box (test3) | Return198 | 7 | |
| Robotic manipulation (Reach) | Robotic-Manipulation reach (train) | Performance33 | 7 | |
| Walker Walk | DMControl-GB color-easy (test) | Avg Episode Return805 | 7 |