Machine learning research: the driving force of Artificial Intelligence

Machine Learning has huge potential and Imperial can be at the forefront of it, say experts as a new initiative launches today.

Dr Marc Deisenroth, from the Department of Computing at Imperial College London, and colleagues from across the College are launching the Machine Learning Initiative today.

Colin Smith caught up with Dr Deisenroth to discover what machine learning is and why it is so important in our daily lives.

Can you tell us more about the launch?

Industry leaders from companies such as Microsoft and Twitter, academics and students from Imperial and representatives from the UK’s major funding bodies, will all be attending.

The Initiative will bring together machine learning researchers from across the College and beyond to provide a collaborative environment for learning, teaching, and research in the field. The aim is to co-ordinate joint activities, such as seminars, workshops, tutorials, summer-schools and grant applications.

What is machine learning?

Machine learning is when algorithms and methodologies give computers the ability to automatically learn and improve from experience without human intervention and without being explicitly programmed. Machine learning automatically finds patterns and structures in data that humans cannot process easily in order to make predictions and decisions. The key emphasis is on “automatic”. No specific human guidance or expert knowledge is required. Machine learning algorithms can automatically adapt to evidence from data, which allows them to learn new concepts.

Is machine learning a form of Artificial Intelligence?

Machine learning can be considered the engine of modern AI. It provides the underlying technology that drives AI. AI is about complex systems that behave intelligently. In order to reach this goal, AI poses many questions, and machine learning provides the technologies toward answering these questions. In other words, AI is about systems and questions whereas machine learning is about practical solutions to these challenges. Another difference is that AI strives for intelligence, whereas machine learning does not necessarily do this.

Why does the public know more about AI than machine learning?

I think it is easier for us to conceptualize an intelligent robot rather than an intelligent algorithm. In this sense, AI is a much more concrete, less abstract concept than machine learning.

Can you give me some examples of machine learning in our daily lives?

Machine learning is the underpinning technology that makes Siri and Alexa work. Google search results are provided by machine learning algorithms. Online retailers provide personalized product recommendations thanks to machine learning. Smart homes use machine learning algorithms to adjust temperatures based on our behaviour. Spotify personalises playlists based on our listening behaviour. When we shoot photos with our digital cameras or phones we often use face detection, again, this is based on machine learning algorithms. Automatic translation systems on our phones use machine learning. We obtain personalised “friend” suggestions on Facebook determined by a machine learning algorithms.

These are only a few examples, but machine learning is already a huge part of our daily lives.

What research is Imperial currently doing in this field?

Our research in this field is broad, but a few examples include:

One of the focuses of machine learning research at the College is on probabilistic modelling. In this area, we use probabilistic models to help us make predictions, based on existing knowledge. For example, when we want to make a statement about how a robot walks on sand without having ever seen the robot walking on sand it is often useful to use probabilistic models.

We can use probabilistic models to describe the spread of infectious diseases, for weather forecasting, high-energy physics, financial applications or the design of experiments.

Another area we focus on is reinforcement learning. This is about learning to solve problems from trial and error. We often think of reinforcement learning resembling the way small children learn. For example, when a child tries something out, he or she receives a reward or punishment from a parent that tells the child whether their attempt was good or bad. The child takes this feedback into account to adjust their actions. At Imperial, we develop novel reinforcement learning algorithms and apply them to robotics or in the healthcare domain.

What do you hope to get out of today’s launch?

We want to display some of Imperial’s potential in machine learning, but also raise the awareness of the importance of the field. There is a lot of enthusiasm and excitement surrounding machine learning. We want to exploit this and raise Imperial’s profile in this field to attract future talent and to become a global player in machine learning research.