A$11.48 – A$41.51

Understanding Autoencoders for Anomaly Detection and Data Generation

Event Information

Share this event

Date and Time

Location

Location

Tutorial room 210

14 Alliance lane

Monash University, Clayton campus

Australia

View Map

Refund Policy

Refund Policy

Refunds up to 7 days before event

Event description

Description

Autoencoders are simple but powerful neural networks that are often used in anomaly detection, data summarisation, and even the generation of new “artificial” but coherent data. This workshop is an introduction to the autoencoder and its capabilities. It is a technical workshop tailored for those who have an introductory understanding of deep learning and want to explore more complex variants of neural networks.

We will start by explaining the structure and key concepts behind the autoencoder, and how these uniquely structured networks can be used to solve many interesting real world problems. You would then proceed to build and train your very own autoencoder from scratch using Tensorflow. We will rigorously analyse the performance of the trained network to inspect its learning capabilities and elaborate on ways the autoencoder can be modified to perform more advanced tasks.

Difficulty: Level 2 - Intermediate

Contents

  • Explanation of the structure and key concepts behind the autoencoder network.
  • Discussion on the potential uses of an autoencoder.
  • Building and training your own autoencoder from scratch.
  • Testing the ability of your autoencoder to perform anomaly detection.
  • Exploring disentangled feature representation leading to variational autoencoders and beta-VAEs.

Itinerary

4:00-4:30 pm – Setup time. This time is for general consultation if you have trouble installing the relevant softwares on your computer. There are no guarantees that we can help solve all problems, so it is best to get things sorted before the day.

4:30 pm – Workshop starts.

6:30 pm – Targeted workshop end time.

6:30-7:00 pm – Extra time for questions and help.

You need

  • A laptop that has Numpy, Tensorflow, Python and any Python code editor installed.
  • An introductory understanding of how neural networks work.
  • Basic knowledge on how to write simple Python code and understand simple Tensorflow syntax. You do not need to be an expert. If you have no relevant knowledge, just spend an hour or two doing online tutorials before the workshop. The rest should be simple enough for you to Google as you go. Or you could attend our introductory workshop “Introduction to Neural Networks and Tensorflow”.
  • A passion for learning, and a ticket.

Does my laptop need to have a GPU? No. This workshop has been carefully tailored to avoid the need for a GPU. The aim here is to explore various concepts and ideas, not necessarily to achieve world class neural network performance.

Join the deep learning community at our Facebook page for event updates, news, and support!

Share with friends

Date and Time

Location

Tutorial room 210

14 Alliance lane

Monash University, Clayton campus

Australia

View Map

Refund Policy

Refunds up to 7 days before event

Save This Event

Event Saved