Students launch Machine Learning Society at Imperial

Two Imperial undergraduate students have launched a new multidisciplinary Machine Learning Society.

Undergraduates Harry Berg (Mechanical Engineering) and Haron Shams (Design Engineering) have set up the Imperial College Machine Learning Society to get students involved in and inspired by technology that’s going to change the world. Here they tell us more about what inspired them, what happened on launch day and their plans for the future.

Image above: Antonia Creswell teaches the audience about the history of machine learning, specifically deep learning.

Why did you set the society up?

Harry: We really wanted to emphasise the interdisciplinary potential of machine learning – it’s not just for computing students, or postgraduates – we’re keen to give everyone, particularly undergrad students, the opportunity to get involved. Everybody can use machine learning.

Haron: In a few years we believe machine learning will be a fundamental skill, like mathematics and programming, so we think it’s really important that students have the opportunity to find a way into this area of research. Previously there was no place specifically for undergrads to go and learn about this technology at Imperial, get involved in projects and meet like-minded people, so we thought we’d set it up ourselves!

What did the launch event involve?

Harry: Our first speaker was Nathan Benaich, founder of London AI, a group, which focuses on trying to help machine learning and artificial intelligence startups grow. Shakir Mohammed, senior research scientist at Deepmind, then spoke about why we want machines to learn and the capability of the kind of systems we can build. PhD student Antonia Crewswell talked about the history of machine learning, which was great because we’re trying to attract people who don’t have a background in this field.

AI has the power to change a lot of things...inclusion, diversity and ethical implications are really fundamental. Harry Berg Mechanical Engineering

Haron: Seth Flaxman then talked about the social impacts machine learning can have, for example forecasting crime and the potential for AI to influence how efficiently we run our societies by helping us to focus our resources. Petar Kormushev, director of Imperial’s Robotics Intelligence Lab, showed the audience demos of the De Niro robot, which is able to walk around campus, order food and play hockey. Marco Marchesi, head of tech at Happy Finish, then explored how machine learning can be used in the creative industry, for example creating 3D scenes from 2D images. 

Harry: Finally Aldo Faisal, Director of the Brain & Behaviour Analytics Lab, spoke about valuable applications of machine learning in medicine, specifically about how it can be used to replace expensive hardware with more cost effective and easily distributed intelligent software, ultimately improving diagnosis and data collection. He also spoke about its potential to help with cognitive diseases such as dementia.

Haron: We got speakers from really varied backgrounds to show how widely machine learning applications can be utilsed and to emphsise how interdisciplinary and valuable this technology can be across the board.

The event proved popular! You mentioned that you welcomed more than 250 attendees…

Haron: One thing we didn’t expect were so many PhD students. Usually societies attract undergraduates, but half the audience were PhD students, and half were undergraduates – it just shows how many students are interested in the field.

What’s lined up for the future?

Harry: The three pillars of our society are tutorials, projects and events. We’ve started teaching both an introductory foundations to machine learning course called Ground Zero, and a more advanced course on deep learning, called In Too Deep Learning. These will cover both core theory and real-time live workshops, where we code together to produce a working algorithm in just one day.

We also just had an event called The Big Brainstorm, a really dynamic session where people explored ideas for projects that the society can work on. Around 60 people attended – from first year undergraduates right up to PhD level students – and we ended up with about eight great ideas from totally varied disciplines that we’d like to run. Anyone who’s interested in getting practically involved should get in touch! The best way to learn is to get out there and do it, so our goal with the project stream is to get people working on an idea, developing skills in a practical way.

Haron: And to make something impressive together, so that people can say, “yeah, we worked on this!”

Harry: The third stream is to hold events, getting external speakers to give either inspirational talks about what their company or group is doing, or to give workshops in their area of expertise.

We want to make sure that the impact of machine learning and AI is positive, and to make people well informed about this technology. Haron Shams Design Engineering

Haron: The events are a chance for us to invite speakers who are at the cutting edge of the field to talk about the latest breakthroughs in machine learning. They’re also an opportunity to explore the different issues around this technology. One of our first events will be about women and diversity in AI – no one should be excluded or feel like they can’t just turn up and get involved. We’ll also be organising events that consider the ethical impacts of AI, to explore how it can be created in such a way that it’s used to improve society rather than be detrimental to it.

Harry: AI has the power to change a lot of things. It’s going to be unavoidable in the future and so inclusion, diversity and ethical implications are really fundamental to consider.

Haron: We want to make sure that the impact of machine learning and AI is positive, and to make people well informed about this technology.

Harry: People have worries about the impact of AI. We want to be open about what AI is, what its limitations are, and how it can be used for good. AI isn’t something that spontaneously turns malevolent - it’s something which simply maximises a utility function.

If you’d like to find out more about the Machine Learning Society check out their Facebook page / Twitter: @ML_ICL / the Student Union website.