LCN seminar: Our Brain plays jazz: Self-Organization of the Liquid State Machine
Event details
| Date | 24.01.2011 |
| Hour | 11:15 |
| Speaker | Gordon PIPA, Institute of Cognitive Science, University of Osnabrueck |
| Location | |
| Category | Conferences - Seminars |
Abstract:
The liquid state (Markram and Maass 2002) and echo state machine (Jäger 2004) ('LSM') had been proposed as promising computational neuronal models. However, the original LSM proposed a random but fixed recurrent network and is therefore incompatible with the observation that biological neuronal networks are constantly changing on many different scales. Therefore, we extended the original idea of the LSM and allowed for self organized changes of neuronal activity and of the network itself. To this end we included two types of plasticity. As a first type we included spike timing dependent plasticity (STDP) that changes the synaptic strength and has been associated with sequence learning and structure formation in recurrent networks. The second type is intrinsic plasticity (IP) that changes the excitability of individual neurons to maintain homeostasis. Based on extensive simulation studies we demonstrate that the combination of both types first optimizes the information processing, second leads to self organized criticality of the network dynamics, and third that the intrinsic noise introduced by intrinsic plasticity increases the robustness of information processing in a high noise regime. We also demonstrate that using these kind of plasticity leads to activity pattern in the system that are compatible with receptive fields or place fields as known form the visual system or the hippocampus. This illustrates that a computational machine using transient activity, such as reservoir computing, is well compatible with well supported coding concepts such as receptive fields.
Acknowledgments:
This work was supported by the grant: EU Grant - Phocus - www.phocus-project.eu
Related Papers:
1. Lazar, G. Pipa, J. Triesch. Fading memory and time series prediction in recurrent networks with different forms of plasticity, Neural Networks 20, 312-322, (2007)
2. A. Lazar, G. Pipa, and J. Triesch, SORN: a self-organizing recurrent neural network, Frontiers Computational Neuroscience 2009, Volume 3 page 23
3. Maass, W., Natschlager, T., & Markram, H.. Real-time computing without stable states: A new framework for neural computation based on perturbations. Neural Computation, 14, 2531-2560, (2002)
Practical information
- General public
- Free