"Spiking Neural Networks for Temporal Pattern Recognition"

Meeting Details

For more information about this meeting, contact Becky Halpenny.

Speaker: Peter Gael, Co-Advisers: Lyle Long & Qiang Du, Penn State

Abstract: My research focusses on the usage of spiking neural networks as a general tool for temporal pattern recognition. Whereas standard sigmoid-type neural networks have had great success in the identification of static patterns, there is some indication that spiking neural networks will have an affinity for the recognition of patterns characterized by a time-evolving sequence of features. A motivating application which I intend to investigate is automatic speech recognition; specifically, the identification of words based on feature sequences extracted from short audio samples. I plan to develop effective coding methodologies and efficient training algorithms in order to evaluate the efficacy of neural networks based on the `theta' spiking neuron for these and other tasks in temporal pattern recognition.

---

Room Reservation Information

Room Number: 105 AG SC IN

Date: 04/26/2011

Time: 9:00am - 11:00am