More than driverless cars: machine learning at University of London
Machine learning is the invisible genie that coordinates your takeaway delivery with no time-consuming human interaction. It is the powerhouse optimising your route from home to the office. It is the force behind the algorithms that enable computers to beat the best human minds in chess. It powers computer systems that can diagnose cancers and systems that can determine complicated financial transactions. It is behind every answer Alexa ever gave you. Machine learning is everywhere.
Although machine learning in essence involves minimal human interaction, the development and implementation of machine learning applications relies on experts in the field of artificial intelligence who have the programming, mathematical and engineering skills required to work with AI systems. University of London and member institution, Goldsmiths, have developed the MSc Data Science programme which offers specialisms in artificial intelligence and financial technology. Dr Jamie Ward is a lecturer in machine learning on the MSc Data Science programme. We spoke to him about the scope of his module, the role of machine learning in our society, and what students from many backgrounds can learn from it.
“We start at the very beginning,” Dr Ward explains. “What is machine learning? How is it used? How is it abused? What are the underlying mathematical principles that enable it to work the way it does?”
The benefits of machine learning are undeniable. It helps us to do things at large-scale that we were never able to do before. Just look at logistics or optimisation - we can now model and predict energy consumption better than ever.
You don’t need to be a maths wizards or have experience coding to be successful, and machine learning can be applied to so many fields that Dr Ward says the programme attracts a diverse range of students.
“We see a lot of artists using it creatively. For example, one art student created an immersive video installation for her final degree project that used video clips pulled from the internet then manipulated them using an algorithm to explore ideas of intimacy. It was a beautiful piece and very impactful. And it used almost all of the key concepts and skills she had developed in machine learning.”
Students can also expect to learn Python, a programming language, on the programme and some of the most popular algorithms being used today. But in addition to teaching fundamental concepts, Dr Ward wants to put machine learning in context for students by exploring its impact on the world.
“The benefits of machine learning are undeniable. It helps us to do things at large-scale that we were never able to do before. Just look at logistics or optimisation - we can now model and predict energy consumption better than ever. By helping to fine-tune our alternative energy sources, it’s transforming wind power in the UK. But for me, machine learning is a tool. Like any tool, it comes with downsides as well as upsides.”
Dr Ward tries to temper some of the hype around artificial intelligence by showing students its limitations and disadvantages too. For example, while we can use it to optimise wind power, there are environmental costs inherent to methods like deep learning. The process of running these high-powered algorithms use a tremendous amount of energy.
“In some cases, training up a neural-network model uses as much energy as flying as a trans-Atlantic 747,” he explains. “As we consider the future of the planet, we need to consider all the variables at play.”
This power is in our hands as programmers. We shouldn’t just be looking for the quickest or most efficient solutions to a problem. It’s important to imagine what the ethical implications might be if we take one route over another.
In addition to environmental ethics, Dr Ward explores questions of data privacy and ownership through the module. He aims to instil a sense of responsibility on the part of the budding programmer by exploring the ethical implications of their actions.
“There’s no doubt of machine learning’s ability to make our lives easier. Voice-activated machines like Alexa and Google Assistant are extremely helpful. But there’s always a trade-off between the ethical issues, especially around privacy, and the benefits we get from this kind of technology. The more a computer is able to understand you- your needs, wants, and desires- the more it is able to help you. But the more invasive that knowledge can be.”
That’s a lesson Dr Ward is keen to pass on to his students.
“This power is in our hands as programmers. We shouldn’t just be looking for the quickest or most efficient solutions to a problem. It’s important to imagine what the ethical implications might be if we take one route over another.
“You read a lot of great research papers in this field that will conclude an exciting study with something along the lines of, ‘and of course there are privacy issues we need to be aware of.’ Privacy should be a priority not an afterthought. In the excitement of research and development, there’s a real lack of attention to this fact that should be built into the design.”
This is one of the main tools of technology at the moment so if you are doing anything from business to public policy, then understanding machine learning and what it’s capable of is a very useful thing.
Dr Ward is currently teaching the first online version of his machine learning module and enjoying the process of translating these lessons to the digital realm. Rather than trying to transpose the classroom experience onto a digital platform, Dr Ward says the module tries to capitalise on the strengths of distance education.
“It’s very different,” he says. “If you imagine campus classes as a live theatre performance, the online module is a carefully edited film. Like a film you don’t have to leave your house to see it, but it has a different quality.
“Normally long lectures are broken into 10-minute chunks. For each of the mini-lectures we’ll have a self-assessment so students can make sure they are on track and understanding the material. We include both essential and recommended reading to encourage students to build their skills and confidence in self-directed learning.”
The main advantage of the online programme, Dr Ward says, is that students can work at their own pace.
“It’s a degree so it still requires a lot of study hours, but you can choose how you spend those hours. It works very well for people who have other commitments and we see this all the time at Goldsmiths.”
He says his virtual classroom is filled with young students with backgrounds from finance to physics, as well as older students looking to supplement their knowledge for their jobs.
“This is one of the main tools of technology at the moment so if you are doing anything from business to public policy, then understanding machine learning and what it’s capable of is a very useful thing.”
Explore how you could apply machine learning to your industry with an MSc Data Science from the University of London.