Power-law forgetting in synapses with metaplasticity
A. Mehta, J.M. Luck

TL;DR
This paper introduces two models of metaplastic synapses demonstrating power-law forgetting, highlighting how different decay mechanisms influence transient responses but share a universal long-term forgetting behavior.
Contribution
The paper develops and compares two novel models of metaplastic synapses with distinct decay mechanisms, revealing universal power-law forgetting behavior.
Findings
Both models exhibit power-law forgetting with the same exponent.
Transient responses differ based on decay mechanisms.
Long-term memory decay follows a universal power-law pattern.
Abstract
The idea of using metaplastic synapses to incorporate the separate storage of long- and short-term memories via an array of hidden states was put forward in the cascade model of Fusi et al. In this paper, we devise and investigate two models of a metaplastic synapse based on these general principles. The main difference between the two models lies in their available mechanisms of decay, when a contrarian event occurs after the build-up of a long-term memory. In one case, this leads to the conversion of the long-term memory to a short-term memory of the opposite kind, while in the other, a long-term memory of the opposite kind may be generated as a result. Appropriately enough, the response of both models to short-term events is not affected by this difference in architecture. On the contrary, the transient response of both models, after long-term memories have been created by the…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
