Science issue: A machine that can learn to speak to you

1b-malica-gasic-photo
News
Have you ever talked to Siri and asked yourself how one builds such a system? Some time ago, when I was pursuing my MPhil degree in Cambridge, Prof. Steve Young demonstrated a spoken dialogue system during a talk. I was fascinated by the idea that one could make a computer speak and understand human speech. I thought I must get into this research and so I applied for a PhD at the Department of Engineering’s Dialogue Systems Group.

A spoken dialogue system normally has three parts: speech understanding, which decodes the meaning from the user’s speech; dialogue management, which tries to come up with a good response; and speech generation, which turns the answer into natural speech. All of these modules can be data-driven: machine learning methods allow us to build systems that become better at their tasks the more data they have.

This is very exciting because in today’s world we are generating data at the biggest pace ever.

There are two distinct kinds of machine learning methods that we use for this research. One is called supervised learning.  This is how we learn ourselves when we have a teacher to provide examples. The system simply tries to imitate the teacher.  Another is called reinforcement learning, and one can think of it as learning from interaction. In this approach, the system can explore different possibilities.  Whenever it makes a good decision, it gets a reward from the user.  Over time, it tries to maximise that reward.  Just like a child learns from trial and error.

This kind of learning through interaction in the context of dialogue systems really intrigues me. The problem is that such learning methods normally need a huge number of interactions before the system starts to behave reasonably well. So I’ve been working on ways to speed up this process, so that the system can learn directly from talking to a human. And indeed I was the first researcher to show that this is possible.

Applications for this technology include every area where we currently see human-computer interaction, and it will make such interaction possible in the future in areas where we can’t imagine it today.  Currently, I am particularly interested in applications in the health sector.  To support such systems, we need to develop algorithms capable of supporting much more complex interactions than what is possible today.  But if successfully built, such systems would have a huge benefit for society.

Dr Milica Gašić
Lecturer in Dialogue Systems, Department of Engineering
Fellow, Murray Edwards College

See my interview for The Naked Scientists: http://www.thenakedscientists.com/HTML/interviews/interview/1001757/

Or check out my website:

http://mi.eng.cam.ac.uk/~mg436/

References

Gašić and S. Young “Gaussian Processes for POMDP-based dialogue manager opimisation”, IEEE Transactions on Audio, Speech and Language Processing, 2014

Gašić, F. Jurcicek, B. Thomson, K. Yu and S. Young. “On-line policy optimisation of spoken dialogue systems via live interaction with human subjects”, ASRU, Hawaii, 2011

Career Path: Blending research and patient care as a GP

1a-fiona-walter-photo
Career
The World Health Organisation defines general practice as providing ‘continuous, comprehensive, co-ordinated and personalised whole-patient care to individuals, families and their communities’. As soon as I completed my clinical training I joined a GP training scheme in Oxfordshire, and have loved being a GP for more than 30 years since then. At its core it’s about being comfortable with being a ‘generalist’ and have some expertise across all clinical conditions, rather than a being a ‘specialist’ with in-depth expertise in one (often very focused) area. When patients first seek help in primary care their problems may be vague or ill-defined- a GP’s expertise lies in working out whether this needs further investigation or referral, or whether the patient can be reassured. One of the most fulfilling parts of being a GP is that we often care for a number of family members over many years. Interestingly, analyses of data from the US, UK and Europe have shown that having more GPs is associated not only with better health outcomes, but also with better patient experience.

After working as a GP partner for more than 12 years, I moved to Cambridge and soon met the newly appointed Foundation Chair of General Practice (Ann Louise Kinmonth, also a New Hall alumna) – she encouraged me to consider a clinical academic career. While continuing to work as a part-time GP, I completed a Masters course and then a doctorate. I was fascinated to find that most of the evidence that we used to care for our primary care patients had arisen from less relevant research from specialist care, and that there was a real need for evidence from the primary care setting.

I now lead the Primary Care Cancer Research group at the University of Cambridge- so, it’s never too late for a mid-career change!

While the career of an academic GP can be demanding, it is also very rewarding. I still work as a GP, but only for one day a week. The rest of the week is spent mainly on research, with some under- and post-graduate teaching. My research focuses on developing patient and GP interventions to help diagnose cancer earlier, as there is plenty of evidence that, for most cancers, a timely diagnosis allows curative treatment and better outcomes. Current projects are researching cancers of the skin, oesophagus, stomach, brain, breast and pancreas. I feel very privileged to work alongside world leaders in cancer screening, early detection and treatment on the Cambridge Biomedical campus, and some of my research findings have already led to changes in NHS guidance for patient care.

What’s next? My research will continue to focus on new and cost-effective approaches for preventing and diagnosing cancer.

One example is the impact of technological advances on patient access to health information, and on the monitoring of symptoms and treatments by both patients and GPs. We need more clinical academics in general practice to take this important work forward.

Dr Fiona Walter MA MD FRCGP
Alumna

Fiona Walter (New Hall 1976) is Principal Researcher (Reader) in Primary Care Cancer Research at the University of Cambridge. She leads studies investigating cancer prevention, diagnosis and follow-up care, was Fellow of Lucy Cavendish College, Cambridge, and is Honorary Clinical Associate Professor at the University of Melbourne, Australia.