The Liquid State Machine is not Robust to Problems in Its Components but Topological Constraints Can Restore Robustness

The Liquid State Machine (LSM) is a method of computing with temporal neurons, which can be used amongst other things for classifying intrinsically temporal data directly unlike standard artificial neural networks. It has also been put forward as a natural model of certain kinds of brain functions. There are two results in this paper: (1) We show that the LSM as normally defined cannot serve as a natural model for brain function. This is because they are very vulnerable to failures in parts of the model. This result is in contrast to work by Maass et al which showed that these models are robust to noise in the input data. (2) We show that specifying certain kinds of topological constraints (such as "small world assumption"), which have been claimed are reasonably plausible biologically, can restore robustness in this sense to LSMs.

• Publication year

  2016

• Venue

  International Joint Conference on Computational Intelligence

• Publication date

  2016-11-30

• Fields of study

  Computer Science

• Identifiers
  DOI 10.5220/0003058902580264
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Structural Properties of the Caenorhabditis elegans Neuronal Network

  2009cited by this paper
• The tempotron: a neuron that learns spike timing–based decisions

  2006cited by this paper
• Small-World Brain Networks

  2006influential reference
• Computational models for generic cortical microcircuits

  2004cited by this paper
• Pattern Recognition in a Bucket

  2003cited by this paper
• Simple model of spiking neurons

  2003influential reference
• Paradigms for Computing with Spiking Neurons

  2002cited by this paper
• Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations

  2002cited by this paper
• Modeling the process of rate selection in neuronal activity.

  2002cited by this paper
• The''echo state''approach to analysing and training recurrent neural networks

  2001cited by this paper
• Competition and multiscaling in evolving networks

  2000cited by this paper
• Topology of evolving networks: local events and universality

  2000cited by this paper
• Emergence of Scaling in Random Networks

  1999cited by this paper
• A logical calculus of the ideas immanent in nervous activity

  1990cited by this paper

• Circuit Optimization Techniques for Efficient Ex-Situ Training of Robust Memristor Based Liquid State Machine

  2022cites this paper
• Computational Efficiency of a Modular Reservoir Network for Image Recognition

  2021influential citation
• Action learning experiments using spiking neural networks and humanoid robots

  2018cites this paper
• Complex Neural Computation with Simple Digital Neurons

  2013cites this paper
• Topological constraints and robustness in liquid state machines

  2012cites this paper
• Learning BOLD Response in fMRI by Reservoir Computing

  2011cites this paper