Dynamic Sparse Training for Deep Reinforcement Learning
This addresses the problem of computational inefficiency in deep reinforcement learning for researchers and practitioners, offering a novel method to accelerate training.
The paper tackles the problem of long training times and high resource consumption in deep reinforcement learning by introducing a dynamic sparse training approach that trains sparse neural networks from scratch and adapts their topology during training. The result is that dynamic sparse agents achieve higher performance than dense methods, reduce parameters and FLOPs by 50%, and reach dense agent performance with 40-50% fewer training steps.
Deep reinforcement learning (DRL) agents are trained through trial-and-error interactions with the environment. This leads to a long training time for dense neural networks to achieve good performance. Hence, prohibitive computation and memory resources are consumed. Recently, learning efficient DRL agents has received increasing attention. Yet, current methods focus on accelerating inference time. In this paper, we introduce for the first time a dynamic sparse training approach for deep reinforcement learning to accelerate the training process. The proposed approach trains a sparse neural network from scratch and dynamically adapts its topology to the changing data distribution during training. Experiments on continuous control tasks show that our dynamic sparse agents achieve higher performance than the equivalent dense methods, reduce the parameter count and floating-point operations (FLOPs) by 50%, and have a faster learning speed that enables reaching the performance of dense agents with 40-50% reduction in the training steps.