Alisa Devlic

James MacGlashan, Evan Archer, Alisa Devlic et al.

Designing reinforcement learning (RL) agents is typically a difficult process that requires numerous design iterations. Learning can fail for a multitude of reasons, and standard RL methods provide too few tools to provide insight into the exact cause. In this paper, we show how to integrate value decomposition into a broad class of actor-critic algorithms and use it to assist in the iterative agent-design process. Value decomposition separates a reward function into distinct components and learns value estimates for each. These value estimates provide insight into an agent's learning and decision-making process and enable new training methods to mitigate common problems. As a demonstration, we introduce SAC-D, a variant of soft actor-critic (SAC) adapted for value decomposition. SAC-D maintains similar performance to SAC, while learning a larger set of value predictions. We also introduce decomposition-based tools that exploit this information, including a new reward influence metric, which measures each reward component's effect on agent decision-making. Using these tools, we provide several demonstrations of decomposition's use in identifying and addressing problems in the design of both environments and agents. Value decomposition is broadly applicable and easy to incorporate into existing algorithms and workflows, making it a powerful tool in an RL practitioner's toolbox.

Xavier Corbillon, Gwendal Simon, Alisa Devlic et al.

The delivery and display of 360-degree videos on Head-Mounted Displays (HMDs) presents many technical challenges. 360-degree videos are ultra high resolution spherical videos, which contain an omnidirectional view of the scene. However only a portion of this scene is displayed on the HMD. Moreover, HMD need to respond in 10 ms to head movements, which prevents the server to send only the displayed video part based on client feedback. To reduce the bandwidth waste, while still providing an immersive experience, a viewport-adaptive 360-degree video streaming system is proposed. The server prepares multiple video representations, which differ not only by their bit-rate, but also by the qualities of different scene regions. The client chooses a representation for the next segment such that its bit-rate fits the available throughput and a full quality region matches its viewing. We investigate the impact of various spherical-to-plane projections and quality arrangements on the video quality displayed to the user, showing that the cube map layout offers the best quality for the given bit-rate budget. An evaluation with a dataset of users navigating 360-degree videos demonstrates that segments need to be short enough to enable frequent view switches.