A Quadratic Actor Network for Model-Free Reinforcement Learning
This is an incremental improvement for continuous control in reinforcement learning, offering enhanced performance with fewer parameters.
The paper tackles the problem of improving model-free reinforcement learning by incorporating quadratic neurons into policy networks, resulting in a 5.8% average increase in reward and 21% better sample efficiency on MuJoCo tasks.
In this work we discuss the incorporation of quadratic neurons into policy networks in the context of model-free actor-critic reinforcement learning. Quadratic neurons admit an explicit quadratic function approximation in contrast to conventional approaches where the the non-linearity is induced by the activation functions. We perform empiric experiments on several MuJoCo continuous control tasks and find that when quadratic neurons are added to MLP policy networks those outperform the baseline MLP whilst admitting a smaller number of parameters. The top returned reward is in average increased by $5.8\%$ while being about $21\%$ more sample efficient. Moreover, it can maintain its advantage against added action and observation noise.