Towards biologically plausible Dreaming and Planning

Humans and animals can learn new skills after practicing for a few hours, while current reinforcement learning algorithms require a large amount of data to achieve good performances. Recent model-based approaches show promising results by reducing the number of necessary interactions with the environment to learn a desirable policy. However, these methods require biological implausible ingredients, such as the detailed storage of older experiences, and long periods of offline learning. The optimal way to learn and exploit word-models is still an open question. Taking inspiration from biology, we suggest that dreaming might be an efficient expedient to use an inner model. We propose a two-module (agent and model) neural network in which "dreaming" (living new experiences in a model-based simulated environment) significantly boosts learning. We also explore "planning", an online alternative to dreaming, that shows comparable performances. Importantly, our model does not require the detailed storage of experiences, and learns online the world-model. This is a key ingredient for biological plausibility and implementability (e.g., in neuromorphic hardware). Our network is composed of spiking neurons, further increasing the energetic efficiency and the plausibility of the model. To our knowledge, there are no previous works proposing biologically plausible model-based reinforcement learning in recurrent spiking networks. Our work is a step toward building efficient neuromorphic systems for autonomous robots, capable of learning new skills in real-world environments. Even when the environment is no longer accessible, the robot optimizes learning by "reasoning" in its own "mind". These approaches are of great relevance when the acquisition from the environment is slow, expensive (robotics) or unsafe (autonomous driving).