Control of synaptic plasticity via the fusion of reinforcement learning and unsupervised learning in neural networks
This addresses a fundamental challenge in neuroscience and AI for researchers, but appears incremental as it combines existing mechanisms.
The paper tackles the Credit Assignment Problem in neural networks by proposing a new learning rule that fuses reinforcement learning and unsupervised learning, resulting in a computational model that minimizes output error through synaptic plasticity dynamics.
The brain can learn to execute a wide variety of tasks quickly and efficiently. Nevertheless, most of the mechanisms that enable us to learn are unclear or incredibly complicated. Recently, considerable efforts have been made in neuroscience and artificial intelligence to understand and model the structure and mechanisms behind the amazing learning capability of the brain. However, in the current understanding of cognitive neuroscience, it is widely accepted that synaptic plasticity plays an essential role in our amazing learning capability. This mechanism is also known as the Credit Assignment Problem (CAP) and is a fundamental challenge in neuroscience and Artificial Intelligence (AI). The observations of neuroscientists clearly confirm the role of two important mechanisms including the error feedback system and unsupervised learning in synaptic plasticity. With this inspiration, a new learning rule is proposed via the fusion of reinforcement learning (RL) and unsupervised learning (UL). In the proposed computational model, the nonlinear optimal control theory is used to resemble the error feedback loop systems and project the output error to neurons membrane potential (neurons state), and an unsupervised learning rule based on neurons membrane potential or neurons activity are utilized to simulate synaptic plasticity dynamics to ensure that the output error is minimized.