Cerebellum Inspired Approach for Pattern Classification in Robots

This project proposes to develop a cerebellum inspired pattern recognition algorithm for robotic data classification. The project aims to investigate the temporal and spatial dynamics in the cerebellar network models capable of predicting cerebellar input-output transformations by analyzing the mathematical and computational properties of the network. Both labs have been working together since 2004 on cerebellar models.

Robotics has expanded to include bio-inspired algorithms for various tasks. Cerebellum has been long known for its role in movement and articulation. CMAC or cerebellar motor articulation control algorithms have existed for more than 35 years although such methods do not faithfully reproduce cerebellar architecture.

The proposal is to exploit biophysical neural network models to the problem of pattern recognition and navigation in mobile robots to achieve practical algorithms for specific applications like surgery or disaster mitigation. Unlike many projects, this project will rely on biological basis for design and function of a pattern classifier that can be used in motor articulation.

Journal papers

• Harilal Parasuram, Bipin Nair, Giovanni Naldi, Egidio D’Angelo, Shyam Diwakar, “A Modeling based Study on the Origin and Nature of Evoked Post-synaptic Local Field Potentials in Granular Layer”, Journal of Physiology-Paris, Volume 105, Issues 1-3, January-June 2011, Pages 71-82, ISSN 0928-4257, 10.1016/j.jphysparis.2011.07.011
• Shyam Diwakar, Paola Lombardo, Sergio Solinas, Giovanni Naldi, Egidio D’Angelo, “Local Field Potential Modeling Predicts Dense Activation in Cerebellar Granule Cells Clusters under LTP and LTD Control”, PLoS ONE 6(7): e21928. doi:10.1371/journal.pone.0021928