The machine is all the rage

Corinna Cortes, Head of Google Research, visited the University of Helsinki as part of the Helsinki Distinguished Lecture Series on Future Information Technology to talk about one of the greatest trend in computing, machine learning.

In New York, Corinna Cortes leads a research team, which she says is “comparable to a small computer science department”, that work on a broad range of problems, including machine learning. Machine learning is a branch of artificial intelligence, which, to put it simply, is about teaching machines to do repetitive tasks without human direction. Machine learning is used in everything from facial and speech recognition technologies to stock market analysis.

No need to learn new languages?

“There is basically no limit to what we can do with machine learning. Say that we make speech recognition even better than it is today. Five years from now we could have this conversation seamlessly in our native tongues while a small device would translate our discussion simultaneously,” says Cortes.

Anyone who has used automated translation software knows that today’s implementations of machine learning are far from perfect, but for every time we use them, the machines get better. Some challenges still need to be solved by researchers, however.

“There is still a lot of work to be done. In speech recognition street noise is a problem, and other people’s voices. When you talk you are hardly ever in a completely quiet environment and it is difficult for a computer to isolate the relevant speech from the noise.”

Computers are blind compared to humans

Speech recognition and machine translation are fairly easy tasks compared to visual recognition, according to Cortes. How can a machine tell a birch apart from a hazel tree, for example?

“Our vocabulary is finite, which means that there are a restricted number of classes that have to be defined. Vision, on the other hand, contains billions of classes that need to be explained for the computer in order for it to be able to recognize an object,” says Cortes.

To be able to make it in the competition-intense job market, Cortes stresses the importance of basic research and learning basic skills, such as algorithms, data structure, statistics, math and programming while still in school.

“You cannot get enough of these basic things. We can teach you all the sexy applications to apply your skills, but if you come up with a good foundation, the job market is going to be much more widely open than if you specialize too much. I would say that 90 per cent of all new hires at Google are hired on basic skills.”

The Helsinki Distinguished Lecture Series on Future Information Technology is a joint project of the University of Helsinki and Aalto University. The focus of the seminar series is to highlight the research challenges and solutions faced by current and future information technology, as seen by the internationally leading experts in the field.