Bio-inspired sequence learning mechanisms and their implementation in a memristive neuromorphic hardware
Bio-inspired sequence learning mechanisms and their implementation in a memristive neuromorphic hardware
We present a sequence learning model that explains how biological networks learn to predict upcoming elements, signal non-anticipated events, and recall sequences in response to a cue signal. The model accounts for anatomical and electrophysiological properties of cortical neuronal circuits and lear...
Saved in:
Personal Name(s): | Bouhadjar, Younes (Corresponding author) |
---|---|
Siegel, Sebastian / Diesmann, Markus / Waser, R. / Neftci, Emre / Wouters, Dirk J. / Tetzlaff, Tom | |
Contributing Institute: |
Neuromorphic Software Eco System; PGI-15 JARA Institut Green IT; PGI-10 Elektronische Materialien; PGI-7 Jara-Institut Brain structure-function relationships; INM-10 Computational and Systems Neuroscience; IAS-6 Computational and Systems Neuroscience; INM-6 |
Imprint: |
2023
|
DOI: |
10.34734/FZJ-2024-01347 |
Conference: | Neuromorphic, Natural and Physical Computing, Hannover (Germany), 2023-10-25 - 2023-10-27 |
Document Type: |
Conference Presentation |
Research Program: |
Human Brain Project Specific Grant Agreement 2 Advanced Computing Architectures Emerging NC Architectures Computational Principles Theory, modelling and simulation Open-Access-Publikationskosten / 2022 - 2024 / Forschungszentrum Jülich (OAPKFZJ) Human Brain Project Specific Grant Agreement 3 |
Link: |
OpenAccess |
Publikationsportal JuSER |
We present a sequence learning model that explains how biological networks learn to predict upcoming elements, signal non-anticipated events, and recall sequences in response to a cue signal. The model accounts for anatomical and electrophysiological properties of cortical neuronal circuits and learns complex sequences in an unsupervised manner using known biological plasticity and homeostatic control mechanisms. We further investigate the feasibility of implementing the sequence learning model on dedicated hardware mimicking brain properties, specifically focusing on memristive crossbar arrays. Finally, we apply the model to sequence classification and anomaly detection in streams of real-world data, and discuss the role of dendritic branches for the sequence learning capacity. |