| Paper: | PS-1B.26 | ||
| Session: | Poster Session 1B | ||
| Location: | H Fläche 1.OG | ||
| Session Time: | Saturday, September 14, 16:30 - 19:30 | ||
| Presentation Time: | Saturday, September 14, 16:30 - 19:30 | ||
| Presentation: | Poster | ||
| Publication: | 2019 Conference on Cognitive Computational Neuroscience, 13-16 September 2019, Berlin, Germany | ||
| Paper Title: | A Model for Perception and Memory | ||
| Manuscript: | Click here to view manuscript | ||
| License: |  This work is licensed under a Creative Commons Attribution 3.0 Unported License. |
||
| DOI: | https://doi.org/10.32470/CCN.2019.1264-0 | ||
| Authors: | Volker Tresp, Sahand Sharifzadeh, Dario Konopatzki, Ludwig Maximilian University of Munich, Germany | ||
| Abstract: | We analyze the close link between perception and memory. Our main hypothesis is that some of the main memory systems of the human brain, e.g., the episodic memory, the semantic memory, and to some degree also the working memory, are by-products of the need for humans to extract more meaningful and more complex information from sensory inputs. Our model is an extension to the tensor memory approach. The key notions are index representations for entities, concepts, relationships and time instances, embeddings associated with the indices, a working memory processing layer, and a sensory memory layer. Perception and memory is realized as an interplay between the different layers. Our model is both competitive to other technical solutions and, as we argue, biologically plausible. Our experiments demonstrate that semantic memory can evolve from perception as a distinguishable functional module. | ||