A Neural Network is a broad term used to represent a vast collection of computational models loosely based on the biological synapses and neurons. Its history dates back to 1940s and since then the field has grown leaps and bounds. Today, Artificial Neural Networks are being widely used in Natural Language Processing, Speech Recognition, Stock Market Analysis, Signal Processing etc.
Inspite of being one of the most exciting parts of Computer Science, there is a vast gap between its research and developer community. This session will attempt at bridging that gap by skipping the non-trivial Mathematics that runs it and providing an operative understanding to the audience, that will be allow them to go ahead and experiment right away.
The session will broadly contain:
1. The type of problems a Neural Network can solve
2. Dissecting the Neural Network structure
3. Walk through of the working of a Neural Network
4. Understanding training
5. Dos and Don’ts while using Neural Networks
6. Demo containing construction of Neural Network using Brain.js
7. Resources and Examples
Inquisitive nature and a Flair for learning something new!
Name: Karthik Hebbar C
Work: Computer Scientist @ Adobe Systems
As far as Neural Networks is concerned, my run-ins with it has resulted in a couple of projects,
1. An NLP to construct a SPARQL query from the user’s search query. This acted as a core for an experimental semantic search engine.
2. A Recurrent Neural Network to predict Surface Roughness based on the operational parameters of a lathe machine.