Memoria: Resolving Fateful Forgetting Problem through Human-Inspired Memory Architecture

Sangjun Park, Jinyeong Bak
2/12/2026

No official code found. Best implementation: cosmoquester/memoria (89 stars, MIT, last commit 2024-10-20).

Abstract

Making neural networks remember over the long term has been a longstanding issue. Although several external memory techniques have been introduced, most focus on retaining recent information in the short term. Regardless of its importance, information tends to be fatefully forgotten over time. We present Memoria, a memory system for artificial neural networks, drawing inspiration from humans and applying various neuroscientific and psychological theories. The experimental results prove the effectiveness of Memoria in the diverse tasks of sorting, language modeling, and classification, surpassing conventional techniques. Engram analysis reveals that Memoria exhibits the primacy, recency, and temporal contiguity effects which are characteristics of human memory.

View on arXivView PDF

Code Implementations(1)

cosmoquester/memoriaCommunity100%
MIT

Memoria is a human-inspired memory architecture for neural networks.

897PythonFeb 15, 20232 years agoMIT
human-inspiredmemorytransformer

Similar Papers

Human-like Forgetting Curves in Deep Neural Networks

86% match2025

Human-inspired Perspectives: A Survey on AI Long-term Memory

84% match2024

Drift to Remember

83% match2024

Teaching AI to Remember: Insights from Brain-Inspired Replay in Continual Learning

83% match1 repo2025

Titans: Learning to Memorize at Test Time

83% match2 repos302024

Towards a Psychology of Machines: Large Language Models Predict Human Memory

83% match2024

Cite this paper

@article{park2026memoria,
  title  = {Memoria: Resolving Fateful Forgetting Problem through Human-Inspired Memory Architecture},
  author = {Sangjun Park and Jinyeong Bak},
  year   = {2026},
  eprint = {2310.03052},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2310.03052},
  journal = {ICML 2023 2023}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2310.03052.svg)](https://codeofpaper.com/paper/2310.03052)

Discussion