Problems with Deep Learning

Speaker image

Siddhartha Rao Kamalakara

ICT Seminar Hall 2

Oct 28, 2018

02:00 PM, 30 minutes

Talk description (Track Two)

Artificial intelligence is probably the fastest evolving field these days and it is so for a good reason. With applications ranging from Medical diagnosis to Quantum physics, AI is redefining the boundaries of computer science. The mathematical models that we use to understand the world around us are flawed and the learning algorithms are very brittle. Thanks to Python and a few frameworks, adoption of AI in the industry is prevalent. But, commercial usage without a proper understanding of what's under the hood has direct consequences and the outcome can get ugly. This talk would elaborate on what the problems with AI are and why we're far from the 'intelligence' part of AI.

About the speaker

Siddhartha is a final year Computer Science student interested in Deep Learning research. He has collaborated with scientists at Google Brain to improve learning mechanisms of neural networks. He also worked with CERN as a Google Summer of Code 2018 student. He loves playing around with new ideas and is here to share.