Event Recordings

Intro to AI & Machine Learning

 

Fundamentals Part 1

Co-president of the Computational Materials Society (CMS) Kevin Mensah will be explaining the fundamentals and background knowledge needed to have a working understanding of machine learning, and its potential use in various fields of materials science & engineering.

Topics Covered:

  • Basic Neural Networks

  • Learning (Supervised/Unsupervised)

  • Classification/Regression

  • Overfitting & Performance Metrics

 

Fundamentals Part 1.5

Co-president of the Computational Materials Society (CMS) Kevin Mensah will be explaining the fundamentals and background knowledge needed to have a working understanding of machine learning, and its potential use in various fields of materials science & engineering.

Topics Covered:

  • Data Science with NumPy library

Fundamentals Part 2

Understanding how data representation and visualization offers insight to how our neural networks perform through their architecture, design, and reactions to training data. It is one of the most essential skills in a machine learning scientists toolkit.

Topics Covered:

  • Data Representation with Pandas

  • Data Visualization with Matplotlib & Seaborn

Fundamentals Part 3

We have finally reached a point where we can design, train and evaluate our very own neural network! This tutorial encompasses the content overed in the Fundamentals tutorials Parts 1-2, to create a convolutional neural network to classify colour images.

Topics Covered:

  • Supervised learning

  • Basic convolutional neural network

  • Error metrics

Fundamentals Part 4

Using a real dataset from the University of California Irvine (UCI), we design, build, and train a neural network to identify if a tumor is malignant or benign based off of 30 different markers. Normally visualizing and correlating a 30-dimensional problem is impossible for humans, but our neural netwrok doesn’t break a sweat with an accuracy of 95%.

Topics Covered:

  • Supervised learning

  • Basic convolutional neural network

  • Data preprocessing of real data

  • Error metrics

Fundamentals Part 5

Following our tutorials on basic neural networks, we'll be taking a look at convolutional neural networks to train AI to examine scientific images. We'll be learning how to train an AI to classify images of malaria cells as parasitic or non-parasitic.

From a materials engineering standpoint, training an AI to recognize patterns in images is extremely useful as much of what we do is correlating observable phenomena to physical properties. This tutorial will be a first step towards using AI to apply to cutting edge materials analysis and discovery work!

Topics Covered:

  • Convolutional Neural Networks

  • Image data pre-processing