THE FUTURE IS HERE

Machine Learning in Python – Session 1. Artificial Neural Networks.

Hosted by the Mechanical Engineering Graduate Students Association, College of Engineering, University of Washington, Seattle.

Speaker(s) :
Sahil Kommalapati (UW MSME’21)
Alrick Cyril Dsouza (UW MSME’21)

Workshop series description:
The four sessions will particularly focus on Artificial Neural Networks, Convolutional Neural Networks, Generative Adversarial Networks and Bayesian Optimization with Hyper-parameter tuning techniques respectively. The primary goal of these workshops would be to expose the applicability of Artificial Intelligence models to our mechanical engineering community and enable immediate integration of these models into current research. These sessions are designed to be collaborative and focus on exercises that are particularly of interest to the mechanical engineering community.

Description for this session:

In the first section on our series, we will introduce Artificial Neural Networks (ANNs). We will go over a brief history of origins, the structure of the network, activation functions, loss functions, mini-batches, and training using back-propagation algorithm. The differences between training, testing and validation datasets are explored. All the executions are done using the Keras Python package, which enables us to easily execute these techniques without writing code from scratch. Towards the end of the session, we present a brief historical review of how ANNs have helped the mechanical engineering community and conclude with an interactive programming session.

Timestamps:
0:00 – Introduction / Overview
4:45 – Introduction to the session

10:55 – Universal Approximation theorem
13:55 – Training neural networks
27:30 – Backpropogation
44:50 – Optimizing Neural Networks
59:30 – ANNs for Mechanical Engineering.
1:10:00 – Tutorial