An STDP-based encoding method for associative and composite data

Abstract

Spike-timing-dependent plasticity(STDP) is a biological process of synaptic modification caused by the difference of firing order and timing between neurons. One of neurodynamical roles of STDP is to form a macroscopic geometrical structure in the neuronal state space in response to a periodic input by Susman et al. (Nat. Commun.10(1), 1–9 2019), Yoon, & Kim. Stdp-based associative memory formation and retrieval. arXiv:2107.02429v2 (2021). In this work, we propose a practical memory model based on STDP which can store and retrieve high dimensional associative data. The model combines STDP dynamics with an encoding scheme for distributed representations and is able to handle multiple composite data in a continuous manner. In the auto-associative memory task where a group of images are continuously streamed to the model, the images are successfully retrieved from an oscillating neural state whenever a proper cue is given. In the second task that deals with semantic memories embedded from sentences, the results show that words can recall multiple sentences simultaneously or one exclusively, depending on their grammatical relations.

Weighted Entropic Associative Memory: A case study on Phonetic Representation and Learning

“…EAM and W-EAM differ from associative memory models developed within sub-symbolic systems, such as the Artificial Neural Networks (ANNs), in the tradition of Hopfield's model 3 and related work [4][5][6][7][8][9][10][11][12][13][14] , in its declarative versus the sub-symbolic format of ANNs approaches; in its constructive versus the reproductive or photographic character of the ANNs models; and in its ability to reject cues not contained in the memory directly without search that the ANNs counterparts lack, among other aspects 2 . However, our concerns have not been the focus of the main line of research in computational models of associative memory, although there are some early relevant exceptions 15 .…”

Implementation of Hebb's rules in a network of excitable chemical cells coupled by pulses

“…In neurobiology, the ability to change a synapse's strength is considered to be the main mechanism by which the phenomena of memory and learning are realized. 16,17 Our previous network with implemented Hebb's rules contains both pulse-and diffusive coupling between excitable cells. 11 Although this network works well, it is difficult to find analytical expressions for boundaries between domains with different dynamic modes.…”

“…In neurobiology, the ability to change a synapse's strength is considered to be the main mechanism by which the phenomena of memory and learning are realized. 16,17…”