A Computational Mechanism for Encoding, Recalling and Forgetting of Real-Life Episodic Events

Episodic memory is one of the crucial dimensions of a human memory system; by imitating them, an agent can achieve human intelligence. A memory of a particular occurrence is referred to as an episodic memory. Due to the fact that every individual has a unique perspective and experience of a certain event, the episodic memories that they have of that event are also unique to each individual. It contains your recollections of individual occurrences, as well as personal information, generic occurrences and pictures of precise times in time you have experienced. To achieve human cognitive abilities, this paper presents a computational mechanism that encodes and forgets the real-life episodic events. The proposed model is based on role entity bindings that encode the real-life conceptual events. Furthermore, the model incorporates a computational gradual forgetting mechanism to achieve higher performance while controlling memory consumption over time. According to the findings of our assessments of the agents, forgetfulness lessens the impact of out-of-date information and states that are not frequently visited on the policies that are generated by the episodic control system. To validate the proposed model, an experimental study is conducted, where the evaluation is done based on event recalling’s hit ratio. The model produces robust performance and got an enhanced hit ratio in event recalling concerning prior methods.