Syllabus

last updated: August 10, 2021

Welcome to the SimCenter ML/AI Workshop 2021. Due to COVID-19, this year’s offering will be an online format. A drawback of an online offering is the lack of interaction among the participants, so we encourage you to be active during the live sessions with questions. The SimCenter intends to create some of the community building - that might otherwise be achieved in-person - through discussion and invites you to join Office Hours with the Development Team to continue these discussions with other members of the Natural Hazards Engineering community.

The bootcamp is organized as a series of pre-recorded video lectures and live online conferencing (Zoom). The lessons and videos are provided in the listing below “by day”.

Warning

Videos must be watched before the online class sessions for the day.

We will spend our live sessions mostly doing hands-on exercises (“In-class Exercise” links below), questions, and discussion. Those exercises are only effective if you did your self-study assignment of watching the videos before we meet.

Warning

To run Exercise 4 on day 1, you need to get a Google Maps API key before the class. Follow this instruction: “Using API Keys” in this link.

• Day 1 - SimCenter Introduction (PDF), Introduction to Machine Learning (Part1-2 and Part3) and BRAILS

  • Pre-class content

    • Introduction to Machine Learning: Video

    • Neural Networks and Deep Learning: Video

    • Image Classification in Tensorflow: Video

    • Warm up with BRAILS: Introduction, Installation (You have to install if you plan to run Exercises 3 and 4 locally on your computer. Alternatively, we also provide notebooks that can run on Google Colab, which requires that you have a Google account. -> Run Google Colab.)

  • In-class Exercises

    • Exercise 1: Conventional Machine Learning: Notebook

    • Exercise 2: Image Classification: Notebook

    • Exercise 3: BRAILS Modules: Notebook

    • Exercise 4: BRAILS City Builder: Notebook

• Day 2 - Object Detection and Semantic Segmentation

  • Pre-class content

    • Sliding Window Object Detection : Video

    • Early One-Stage and Two-Stage Object Detector Algorithms: Video

    • State of the Art Object Detection Algorithms & Detection Datasets: Video

    • An Overview of Image Segmentation Algorithms: Video

    • State of the Art Image Segmentation Algorithms, Segmentation Datasets: Video

  • In-class Exercises

    • Exercise 1: Roof Detection through Semantic Segmentation: Notebook

    • Exercise 2: Crack Segmentation: Notebook

• Day 3 - Neural Network Architectures

  • Pre-class content

    • Multi-Layer Perceptron: Video

    • Convolutional Neural Networks: Video

    • Attention Networks: Video

    • Transformers: Video

  • Demos

    • Demo 1: Hand-written digit classification with single-layer perceptron

    • Demo 2: Hand-written digit classification with convolutional neural networks

    • Demo 3: Hand-written digit classification with attention networks

    • Demo 4: Hand-written digit classification with transformer

  • In-class Exercises

    • Exercise 1: Hand-written digit classification with multi-layer perceptron

    • Exercise 2: Roof classification with Transformation Learning