Mistake Gating Leads to Energy and Memory Efficient Continual Learning
Neural Computation · 05.11.2026 · 1-21
DOI: 10.1162/neco.a.1602 ↗Kurz gefasst
Fachartikel in der Zeitschrift Neural Computation, online erschienen am 05.11.2026. Verfasst von Aaron Pache und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Synaptic plasticity is metabolically expensive, yet animals continuously update their internal models without exhausting energy reserves. However, when artificial neural networks are trained, the network parameters are typically updated on every sample that is presented, even if the sample was classified correctly. Inspired by the human negativity bias and error-related negativity, we propose memorized mistake-gated learning, a biologically plausible plasticity rule where synaptic updates are strictly gated by current and past classification errors. This reduces the number of updates the network needs to make by 50% to 80%. Mistake gating is particularly well suited for incremental learning where new knowledge is acquired on a background of preexisting knowledge and for online learning scenarios when data need to be stored for later replay. Here, mistake gating strongly reduces the required storage buffer size. Moreover, mistake gating increases robustness when learning from imbalanced classes. The algorithm can be implemented in a few lines of code, adds no hyperparameters, and comes at negligible computational overhead. Learning on mistakes is an energy efficient and biologically relevant modification to commonly used learning rules that is well suited for continual learning.
Zitieren
APA
Pache, A., & Rossum, M. C. W. V. (2026). Mistake Gating Leads to Energy and Memory Efficient Continual Learning. Neural Computation, 1-21. https://doi.org/10.1162/neco.a.1602
Vancouver
Pache A, Rossum MCWV. Mistake Gating Leads to Energy and Memory Efficient Continual Learning. Neural Computation. 2026:1-21. doi:10.1162/neco.a.1602
BibTeX
@article{Pache2026,
author = {Pache, Aaron and Rossum, Mark C. W. van},
title = {Mistake Gating Leads to Energy and Memory Efficient Continual Learning},
journal = {Neural Computation},
year = {2026},
pages = {1--21},
doi = {10.1162/neco.a.1602}
}
EUVIMED