## Women in STEM: Dr Anita Faul

*Dr Anita Faul is a Teaching Associate at the Cavendish Laboratory and a Fellow of Selwyn College, where she specialises in algorithms. Here, she tells us about what it's like to teach at Cambridge and whether we can trust the answers that computers give us. *

I think the most fun I’ve probably had at work was when I programmed a movable camera to follow me around the room. I’m a mathematician by training and now work as a Teaching Associate in Scientific Computing, specialising in algorithms. I will soon be starting at the British Antarctic Survey as a Data Scientist, to which I am immensely looking forward to.

Artificial Intelligence and Machine Learning are very popular now. These are also algorithms, with the difference that often the numbers are interpreted as probabilities. So computers do not necessarily give an exact answer, but the answer that is the most probable in some setting. Computer vision has developed a lot in recent years. I've also worked in industry on various applications and particularly enjoy making connections between different fields. The challenge is to express the problem in mathematical terms. Then it can be tackled by algorithms.

With human learning, experiences change how we interpret our world. A levitation act will not fascinate a small child if it has not learned about gravity yet. Once it knows about gravity, it does seem to like throwing things down again and again, as any frustrated parent will tell you! Similarly, machine learning lets the computer have experiences in the form of data - lots and lots of data. While a human child can distinguish between a cat and a dog after seeing a few examples, a computer needs far more.

The most important question is not how a computer arrives at a result, but why. Deep neural networks have had great success lately. However, their structure is so complex that a human cannot understand how they arrived at their answer. How can we then trust the answer? This can also lead to computers being easily fooled where a human would not be. This is something else that we don’t yet understand why. I'm interested in developing algorithms which are self-improving, learning from new data.

The students are my teachers. They ask interesting, challenging questions. It is best to be open, if I do not know the answer, and go on a journey of discovery together. I might not know it, but I surely will find out. Students learn in different ways and I enjoy the challenge to find ways to make a topic accessible. Artificial intelligence makes the headlines often enough to be able to remain topical.

Collaborations are easy if one is willing. A lot of high tech companies working in this field have settled in Cambridge or have opened offices here. Additionally, exciting research is conducted in many departments across Cambridge using machine learning techniques. I enjoy pointing these out to the students who can then see what they have learned in action.

Have a go, you never know what you might achieve. When I was 15, I took part in a maths competition aimed at pupils two years above me at school, since my brother took part. I placed higher than him. He bore it gracefully. For me, it was a start to more and more opportunities opening up. If you do not try, you cannot succeed. Yes, there is failure, but then one readjusts and carries on. Lately, I have become more interested in post-graduate education in general, policies and procedures, funding and finances. The information is too dispersed, especially for those considering a post-graduate degree. I'm working on linking different sources of information.