Encoding and Recall of Spatio-Temporal Episodic Memory in Real Time

Abstract

Episodic memory enables a cognitive system to improve its performance by reflecting upon past events. In this paper, we propose a computational model called STEM for encoding and recall of episodic events together with the associated contextual information in real time. Based on a class of self-organizing neural networks, STEM is designed to learn memory chunks or cognitive nodes, each encoding a set of co-occurring multi-modal activity patterns across multiple pattern channels. We present algorithms for recall of events based on partial and inexact input patterns. Our empirical results based on a public domain data set show that STEM displays a high level of efficiency and robustness in encoding and retrieval with both partial and noisy search cues when compared with a stateof-the-art associative memory model.

Who Are the `Silent Spreaders'?: Contact Tracing in Spatio-Temporal Memory Models

“…Previous models of episodic memory include Episodic Memory-ART (EM-ART) [16], [17] and STEM [11]. While EM-ART learns a hierarchical encoding scheme of sequential events and episodes, STEM explicitly represents the time and space of individual events without an episode layer.…”

“…Time complexity of STEM-COVID: In the standard fusion ART model, the worst-case time complexity for learning a category node is O(mn 2 ) if no matched node is found, where m is the number of attributes in the input fields and n is the number of existing category nodes [11]. The quadratic component is incurred by the repeated code competition and template matching for resonance search.…”

“…It enables a cognitive system to develop adaptive behaviors by recalling past experiences. For an efficient encoding and recalling mechanism, this work extends a Spatio-Temporal Episodic Memory (STEM) model [11] to encode the spatiotemporal traces and the tested positivity of SARS-CoV-2 of different individuals. The building block of the model is a class of biologically-inspired self-organized neural networks, known as fusion Adaptive Resonance Theory (ART) networks [12], [13].…”

An Autonomous Developmental Cognitive Architecture Based on Incremental Associative Neural Network With Dynamic Audiovisual Fusion

“…That means high-level information cannot feedback to low-level networks. Chang and Tan [40] and Chin et al [41] based on ART only used input to activate a cognitive node in the category field and read out the weight of the winner node. Kasaei et al [42] based on hierarchical object representation and extended Latent Dirichlet Allocation model also focused on the classification task and pre-defined many learning parameters.…”