Neural Networks and TensorFlow: A Gentle Introduction
Date:
This talk was given as part of Milan Curcic’s Lunch Bytes seminar series at RSMAS. Lunch Bytes seminars are an informal way for students, faculty, and staff to share their computing-related knowledge with other members of the RSMAS community.
Abstract: This seminar will provide an introduction to the basics of both TensorFlow and neural networks in general. Much of the mathematics of neural networks will be avoided, as the focus is on helping interested students and researchers determine if deep learning is a good fit for their current or future work rather than the nuts and bolts of deep learning. Some questions I hope to answer are: What’s behind all the hype around deep learning? What does my output mean? What types of network fit what type of problem? The other focus of this seminar is the TensorFlow library for deep learning (among other things). I hope to answer questions about TensorFlow such as: What are the benefits? How can I visualize my network performance? Can TensorFlow speed up my development time?