A brain-inspired associative memory model with olfactory context for visual scene recall
Nonlinear Theory and Its Applications, IEICE · 2026
DOI: 10.1587/nolta.2026nhp0010 ↗Kurz gefasst
Fachartikel in der Zeitschrift Nonlinear Theory and Its Applications, IEICE, 2026. Verfasst von Haruhiko Watanabe und 3 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben.
Abstract (Originalsprache)
This study proposes a brain-inspired associative memory model that incorporates olfactory information as contextual input and uses a modern Hopfield network (MHN) for rapid storage of episode-specific traces. The model combines a multimodal variational autoencoder with the MHN to integrate visual and olfactory inputs and to retrieve stored episodes from partial cues. Experiments used synthetic visual scenes and odor signals measured by an electronic nose sensor. Odor-only input recalled scene contexts associated with each odor condition. Under partial visual observation, the proposed model achieved higher background-level recall accuracy than a baseline without olfactory input. In addition, the model recalled newly presented image–odor pairs after a single memory update. These results support the feasibility of using olfactory context for cue-based visual scene recall and rapid episode-specific storage in a brain-inspired associative memory framework.
Zitieren
APA
Watanabe, H., Mizutani, A., Tamukoh, H., & Tanaka, Y. (2026). A brain-inspired associative memory model with olfactory context for visual scene recall. Nonlinear Theory and Its Applications, IEICE. https://doi.org/10.1587/nolta.2026nhp0010
Vancouver
Watanabe H, Mizutani A, Tamukoh H, Tanaka Y. A brain-inspired associative memory model with olfactory context for visual scene recall. Nonlinear Theory and Its Applications, IEICE. 2026. doi:10.1587/nolta.2026nhp0010
BibTeX
@article{Watanabe2026,
author = {Watanabe, Haruhiko and Mizutani, Akinobu and Tamukoh, Hakaru and Tanaka, Yuichiro},
title = {A brain-inspired associative memory model with olfactory context for visual scene recall},
journal = {Nonlinear Theory and Its Applications, IEICE},
year = {2026},
doi = {10.1587/nolta.2026nhp0010}
}
EUVIMED