Skip to ContentSkip to Navigation
Research Zernike (ZIAM) Bio-inspired Circuits & Systems Chicca group

Unsupervised learning of spatio-temporal patterns with self-organizing spiking neural networks

Type and duration:

Master project/thesis, flexible duration.

Context:

Local synaptic plasticity is the key mechanism of cortical plasticity which enables self-organization in the brain, that in turn enables the emergence of consistent representations of the world. Self-organization can be defined as a global order emerging from local interaction without an external supervisor. Thus, in addition to locality, it is unsupervised by nature and radically contrasts with supervised learning using labeled data. Recent works have explored the potential of self-organizing neural networks with local plasticity mechanisms for unsupervised learning [1] [2] and multimodal association [3], but the proposed SNN topologies only learn the underlying spatial structure of the patterns and do not capture the inherent temporal information that are necessary to recognize patterns where timing matters [4], such as speech. In this project, we want to explore several local plasticity mechanisms as well as SNN topologies for unsupervised learning of spoken words.

Objectives:

  • Literature review of the previous works for unsupervised learning of spatio-temporal patterns with spiking neural networks.
  • Propose a new topology of recurrent SNNs to be able to learn the temporal information in the input spike pattern without using labels.
  • Explore and confront different synaptic plasticity rules using the BindsNET simulator [5].

Required skills:

Python programming, basic machine learning knowledge, strong interest in computational and cognitive neurosciences.

Contact person:

Lyes Khacef

References:

  1. Diehl, P.; Cook, M. Unsupervised learning of digit recognition using spike-timing-dependent plasticity. Front. Comput. Neurosci. 2015.
  2. Unsupervised learning of event-based image recordings using spike-timing-dependent plasticity, Laxmi R. Iyer, 2017.
  3. Rathi, N.; Roy, K. STDP-Based Unsupervised Multimodal Learning With Cross-Modal Processing in Spiking Neural Network. IEEE Trans. Emerg. Top. Comput. Intell. 2018.
  4. Is Neuromorphic MNIST Neuromorphic? Analyzing the Discriminative Power of Neuromorphic Datasets in the Time Domain, Laxmi R. Iyer, 2021.
  5. BindsNET, https://github.com/BindsNET/bindsnet
Last modified:09 September 2021 3.50 p.m.