Recent technological advances in micro and nano-fabrication technology and high-yield electrophysiology techniques allowed us to record the activity of hundreds/thousands of neurons simultaneously. This has spurred renewed interest in applying multi-electrode extracellular electrophysiology approaches in the field of neuroscience. Each electrode samples the activity of one or more neurons in its vicinity. One of the major challenges is to efficiently and robustly detect the spikes that individual neurons fire from the raw recorded electrophysiological signals. Current methods are computationally demanding, slow, unreliable in noise rejection, sensitive to optimal selection of parameters and usually require human supervision. The aim of this project is to develop a new spike sorting approach using machine learning and deep learning methods. The desire method will identify spikes, remove artefacts and noise from raw data including photoelectric artefacts from optogenetics and imaging or motion artefacts from cables and movement of animals, and predict the occurrence of spikes based on other electrophysiological measures. Direct work with animals is not required, however, if interested, there will be a unique opportunity to observe or contribute in animal experiments.
C++/Matlab or Python, basic knowledge of machine learning