DL RTP Guest- Dr. Semework "Using AI techniques to predict brain activity"

2018-04-11
Organizer: Deep Learning RTP
Click here for registration info

Dr. Mulugeta is an associate research scientist at CUMC currently studying how the brain remembers objects in our familiar environment. He records from the brain and sees how this memory works when we don't need it for what we are doing now. He has used artificial intelligence approaches to predict future memory performance for neurons whose activity is recorded when a subject does non-memory tasks. He has successfully implemented various Deep Learning (TensorFlow) models to predict future neuronal performance. Since there was a limitation in data size, he used augmentation techniques to increase the size and create artificially perturbed inputs (such as translated, warped, etc.) to mimic naturalistic inconsistencies in data. His findings indicate that tools such Deep Learning can be used to correctly forecast the future.

----

Poster: triangletech