Course on Machine Learning with Applications in Media Engineering
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
handouts
keynotes
misc
README.md
_config.yml

README.md

EE4C16

[EE4C16] is an introduction course to Machine Learning (ML), with a focus on Deep Learning. It is a fourth year module offered by the Electronic & Electrical Engineering department to the undergraduate students of Trinity College Dublin.

Although Deep Learning has been around for quite a while, it has recently become a disruptive technology that has been unexpectedly taking over operations of technology companies around the world and disrupting all aspects of society. When you read or hear about AI or machine Learning successes in the news, it really means Deep Learning successes.

The course starts with an introduction to some essential aspects of Machine Learning, including Least Squares, Logistic Regression and a quick overview of some popular classification techniques.

Then the course dives into the fundamentals of Neural Nets, including Feed Forward Neural Nets, Convolution Neural Nets and Recurrent Neural Nets.

The material is constructed in collaboration with leading industrial practitioners including Google, YouTube and Movidius, and students will have guest lectures from these companies.

Labs

We have designed a unique environment specifically for this course so that students can learn best industry practices.

Our web platform can transparently connect students to a Google Cloud Platform cluster via web based terminal/editor/Jupyter sessions. Labs use the Keras framework and are automatically assessed using Git to give immediate feedback.

Labs include designing and training various DNN for image classification challenges, self driving car (simulator) and text processing.

Handouts

labs

00 - Introduction

01 - Least Squares/Linear Regression

02 - Logistic Regression

03 - Classic Classifiers

04 - Evaluating Classifier Performance

05 - Deep FeedForward Networks

06 - Convolutional Neural Networks

07 - Advances in Neural Networks Architectures

08 - Working with Text

  • pdf slides to come

09 - Recurrent Neural Networks

  • use pdf from Fei-Fei Li & Justin Johnson & Serena Yeung's course at Stanford [https://goo.gl/V7NDbf]. Look at material covered from slide 10 to 76. You also need to know that LSTM and GRU can be used as replacement to simple RNNs. LSTM and GRU were designed to specifically address the problem of vanishing/exploding gradients.

Keynotes

Michaela Blott - Xilinx

Dr. George Toderici - Google Research

  • see BlackBoard

Dr. Sofiane Yous - Intel/Movidius

  • see BlackBoard

Sample Exam