Search

Although transfer learning is considered to be a milestone in deep reinforcement learning, the mechanisms behind it are still poorly understood. In particular, predicting if knowledge can be transferred between two given tasks is still an unresolved problem. In this work, we explore the use of network distillation as a feature extraction method to better understand the context in which transfer can occur. Notably, we show that distillation does not prevent knowledge transfer, including when transferring from multiple tasks to a new one, and we compare these results with transfer without prior distillation. We focus our work on the Atari benchmark due to the variability between different games, but also to their similarities in terms of visual features., Comment: Published at the IEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning (IEEE ADPRL), 2022

International audience; While hardware implementations allow the production of highly efficient and performance oriented designs, exploiting features such as parallelization, their longer time to code and implement often bottlenecks rapid prototyping. On the other hand, high-level synthesis (HLS) tools allow for faster experimentation of software code to a hard- ware platform while demonstrating a reasonable extrapolation of the expected hardware behavior. In this work, we attempt to show a rapid, fast prototyping of the well known HQC algorithm, using HLS, and show how with a modification of certain parameters, varying degrees of comparable results can be obtained. These results, in turn, could be used as a guide for HDL-RTL developers to enhance their designs and better prototyping time in the future. Additionally, we also demonstrate that it is possible to benefit from HQC’s versatility; by achieving a low hardware footprint whilst also maintaining good performances, even on low-cost FPGA devices, which we demonstrate on the well known Artix-7 xc7a100t-ftg256-1.

Learning a good state representation is a critical skill when dealing with multiple tasks in Reinforcement Learning as it allows for transfer and better generalization between tasks. However, defining what constitute a useful representation is far from simple and there is so far no standard method to find such an encoding. In this paper, we argue that distillation -- a process that aims at imitating a set of given policies with a single neural network -- can be used to learn a state representation displaying favorable characteristics. In this regard, we define three criteria that measure desirable features of a state encoding: the ability to select important variables in the input space, the ability to efficiently separate states according to their corresponding optimal action, and the robustness of the state encoding on new tasks. We first evaluate these criteria and verify the contribution of distillation on state representation on a toy environment based on the standard inverted pendulum problem, before extending our analysis on more complex visual tasks from the Atari and Procgen benchmarks., Comment: Published at the 1st Conference on Lifelong Learning Agents (CoLLAs), 2022