This two-day course aims to develop the technical skills necessary for building systems that use machine learning to make automated decisions whilst accounting for ethical objectives. The course includes presentations, discussions, a group project, and hands-on interactive exercise modules which will allow you to consolidate the new concepts from the course.

Outcomes

Having explored example systems that use machine learning to make automated decisions whilst accounting for ethical objectives, you will have gained an understanding of the technical pitfalls that prevent machine learning systems from behaving ethically, as well as how to identify and correct for them.

Prerequisites

This course is for people who have experience building data-driven models; and, are comfortable building statistical models, reading equations and chatting about terms such as “parameter optimisation”, “overfitting” and “model validation”.

Outline of Topics

Automated decision making: We provide a review of the foundations of machine learning and model validation, with an emphasis on ensuring a strong conceptual understanding and the ethical implications of algorithmic decision making.
Loss functions and robust modeling: We investigate the choice of loss function, including what considerations to omit and include, as the primary mechanism of control designers have over the ethical operation of the system. We examine the design choices and assumptions such as encoding values in loss functions, cost-sensitive classification, calibration, and decision making based on predicted probabilities and dataset shift.
Causal versus predictive models: Failure to consider causality can lead to poor consequences despite good intentions. We clarify the distinction between causal and predictive models and how they can be used & interpreted: identifying when a causal model is required and understanding Simpson’s Paradox.
Fair machine learning: We examine some of the common notions of algorithmic fairness that attempt to measure and correct for such disparate treatment or outcome in ML systems.
Interpretability, transparency & accountability: We provide an introduction to some of the tools and techniques available for making models more interpretable and transparent and discuss how to communicate key information about model behaviour and ethical risks to those ultimately accountable for the system.
An applied project that will give you the opportunity to put the concepts learned into practice. During the project, you will work in teams to analyse an algorithmic system, identify potential ethical issues, propose solutions, and present the results to the group at the end of the day.

Instructors

This course is run in groups of up to fifteen students with two Gradient Institute instructors present throughout the course to lead the course and answer any questions. The instructors are members of our team of machine learning practitioners all with more than a decade of experience designing and implementing consequential AI systems. Find out more about our team on our
people page.

Remote delivery

To ensure everyone’s safety, we are currently delivering our courses via remote access to respect social distancing measures. We look forward to being able to offer courses face-to-face once it is safe to do so.

We’ve formatted our online courses to replicate a classroom. We use video-conferencing for our presentation and linked break-out rooms for project work. You’ll be able to interact with our instructors and ask them questions through this and via an audio and text chat app, so that you can do the same things you would in face-to-face, such as ask questions and get to know your fellow students. You will access exercises, interactive models, and visualisations in a Jupyter notebook.

To participate in the course, you will need:

A reliable computer with a webcam, microphone and headphone/speakers
Reliable internet access

More information about systems requirements will be sent out closer to the course.

If you have any questions or to discuss a corporate course for your organisation, send an email to training@gradientinstitute.org

The essentials

This course is for you if you can (or do) build data-driven models professionally;
You’ll need to have demonstrated expertise in a data-driven modeling discipline, either machine learning or statistics to participate;
By the end of it, you will have explored some of the technical pitfalls that prevent machine learning systems from behaving ethically, as well as learning how to identify and correct for them;
Each course has a participant limit of 15 students;
Two instructors are present throughout the whole course to deliver the content and provide support;
This two-day course is delivered remotely between 9:00 am to 5:00 pm Australian Eastern Standard Time (AEST).

Visit our website for more information about our courses and our organisation.

Tickets

We are offering this course at an introductory rate.

Save 10% when you book for yourself and one or more colleagues, and save 10% with our early bird discount. Listed prices include these discounts and are inclusive of GST.

Registration will close on 18/09/2020 at 11:55 pm Australian Eastern Standard Time.