1 July 2019

Newly appointed professor is playing a significant role in the artificial intelligence revolution

New professor

In 2018, Christina Lioma became leader of the Machine Learning research section at the Department of Computer Science, University of Copenhagen (DIKU). On 7 June 2019, she became the second woman to obtain a higher doctorate degree in computer science in Denmark*. And now, on 1 July 2019, she is officially appointed full professor at DIKU.

Christina Lioma at her doctoral defence on 7 June 2019.
Christina Lioma at her doctoral defence on 7 June 2019.

- I’m very happy and proud to be one of the first women to receive a higher doctorate degree in computer science in Denmark. Unfortunately, this also emphasises the gender gap in computer science. But I hope, I can help inspire talented young women to pursue a path in this exciting field and become the next generation of leaders in research and industry, says Christina Lioma.

What has brought Christina Lioma to where she is today is a particularly successful research career with notable breakthroughs within areas such as artificial intelligence (AI), information retrieval, web data mining and natural language processing, as well as a critical viewpoint and scientific curiosity: 

- We as scientists should preserve our critical viewpoint and scientific curiosity. There are plenty of worthy challenges and we should use our research creativity to try and solve these. And this means asking the right questions. When we look at impressive demos from the tech giants, we should not think: “Google has solved everything, I might as well go home”. We should tell ourselves: “Look, with all these resources, and they still cannot get it right! I can do that in my research!”, says Christina Lioma.

And this is exactly what she has been doing in the last 15 years.

In 2018, her research team won the second place (out of 686 submissions) in a Spotify competition on predicting if users will skip or listen to the music they stream. In another international competition in 2018 on fake news detection, her team won the first prize in automatically identifying which parts of a political speech are true or false, and in 2019 they won the first prize in identifying which parts of a political speech should be fact-checked.

Most recently, her team developed a way for machines to read and understand text, as accurately but 43% faster per second than published results by Google, 90% faster than Academia Sinica (the national academy of the Republic of China), and 33% faster than the Allen Institute of AI.

Vision: Creating successful communication between machines and humans

As humans, we have been learning how to live and work with each other for millennia. We have evolved and we have improved through experience, trial and error, how to communicate with each other. But AI has not been around for millennia. It is a newcomer, but it is learning tremendously fast. And it is catching up.

- We are witnessing an AI revolution. As computer scientists, we are first-hand witnesses to this paradigm shift towards increasingly more data-driven and end-to-end autonomous computations. But also outside the computer science field, society at large is experiencing this AI revolution. More and more of our daily interactions are now computer-mediated, or even computer-generated, says Christina Lioma.

This requires a combination of highly advanced technologies that have been developed for a very long time across many different areas, from text to speech synthesis and machine learning, to super-fast algorithms and human-computer interaction. All these advances are combined to produce machines that can (a) understand what humans tell them, and (b) give answers that can be understood by humans.

- The overall effect of this AI revolution is that of a highly disruptive and pervasive technology blending in with how we humans communicate and interact with each other. When we call a customer service and talk to a chatbot instead of a human, or when we issue voice commands to our mobile phones or car navigators. There are very few aspects of our daily lives that are no longer affected by AI, and it is projected that this will increase rapidly in the near future. Therefore, top quality university research in this area is extremely important, says Christina Lioma and continues:

- We are beginning to see some very impressive examples from industry, e.g. Google Duplex (a new technology for conducting natural conversations to carry out “real world” tasks over the phone, like scheduling appointments). However, we should bear in mind that, for every single such example of successful communication between machines and humans, we can easily find at least two examples where this communication fails miserably.

In the period from 2009 - 2017, Christina Lioma worked on this challenge by improving how machines “understand” text, and bringing them closer to how humans process information. This work formed the basis of her higher doctorate dissertation, the first ever in Denmark to be given to a woman. 

- We are at a stage now, where we need to improve technology further, for instance to train it to become more realistic and ethical. I am sure that in the near future we will experience some really good technology, Christina Lioma concludes.

* The higher doctorate degree is called disputats in Danish and habilitation in German and French.