Friday 4:30 p.m.–5 p.m.
Neural Nets for Newbies
- Audience level:
Neural networks have regained popularity in the last decade, but they get dismissed as being too complicated to understand and implement. This talk breaks down the neural net structure as simply as possible, so you have a framework on which to grow your knowledge in the space. I will put neural nets in the context of real-world applications and share Python packages and code where you can get started building your own. Coming out this talk you won't know everything about neural nets, but you will walk away with a solid foundation and some resources on where to go next to learn more.
"Neural Nets for Newbies" is geared to provide clarity on what neural networks are, how to start using them and why they are valuable -- feature engineering. This talk is targeted to anyone who is passionate about understanding algorithms and code to define and leverage patterns in data. Neural network algorithms have a wide range of applications that handle complex and tough-to-model data sets. For example, they are extremely popular in image classification (e.g. identifying people, animals or other objects in photos) and were part of the Netflix challenge solution. In the talk, I will cover a high-level history and breakdown the basic structure of a neural net. I will give an understanding on what to research further to implement a neural net in the real world. I'll list a variety of Python packages that you can use, and I’ll show example Python code for running a simple neural net on your own based on the MNIST dataset (e.g. "hello world" of neural nets). Last, I’ll briefly share where neural nets going in use and impact expectations.