Uber’s Chief Scientist delivers 2018 Booth Memorial Lecture at Birkbeck

Professor Zoubin Ghahramani FRS discussed the theoretical foundations and future directions of machine learning and artificial intelligence.

L-R: Dr Alessandro Provetti, Professor Alexandra Poulovassilis, Professor Zoubin Ghahramani, Professor Philip Powell, Professor Mark Levene

On 12 June 2018, Birkbeck’s School of Business, Economics and Informatics hosted the 2018 Andrew and Kathleen Booth Memorial Lecture, where Zoubin Ghahramani FRS, Professor of Information Engineering at the University of Cambridge and Chief Scientist at Uber, delivered a lecture on the history of machine learning and recent advances in artificial intelligence. 

The Andrew and Kathleen Booth Memorial Lecture is delivered annually by world-renowned scholars and practitioners of computer science, to commemorate the pioneering work of Professor Andrew Booth and Kathleen Booth (née Britten) at Birkbeck. In 1947, both academics visited the Institute of Advanced Studies at Princeton University, where they worked with John von Neumann, one of the most influential early computer pioneers. Their best-known machine, APEC (All-Purpose Electronic Computer), was designed in Birkbeck’s Computation Laboratory in 1952.

It is an exciting time for the fields of computer science and information systems, as artificial intelligence (AI) and machine learning continue to deliver major breakthroughs across industries and show great potential for impactful applications. In his introduction, Dr Alessandro Provetti, Reader in the Department of Computer Science and Information Systems and Director of Birkbeck’s Institute for Data Analytics (BIDA), noted that machine learning is a vast topic and Professor Ghahramani’s contributions to the Bayesian approach in the field are “a gift.”

Professor Ghahramani’s lecture explored the historical foundations of machine learning and deep learning, which is a revival of the concept of neural networks; recent advances in artificial intelligence, especially in the games industry; as well as probabilistic programming and Bayesian optimisation. He also discussed the limitations of deep learning, saying that while it has been instrumental to many innovations, deep learning is data hungry, compute-intensive, and sometimes poor at representing uncertainty: “We can’t base our technology on deep learning alone. We need systems that know when they don’t know, and one way of achieving this is Probabilistic Machine Learning, based on the mathematics of Reverend Thomas Bayes who came up with Bayes’ Rule.”

According to Professor Ghahramani, Bayes’ Rule enables us to better understand the process of learning and ultimately overcome some of the limitations of deep learning: “Bayes’ Rule shows us how to update our beliefs about hypotheses in consideration of data, and it is the process of understanding data – how we go from prior knowledge before observing data, to posterior knowledge after observing data. Probabilities matter for artificial intelligence.”

Concluding with an overview of the future directions of artificial intelligence at Uber, Professor Ghahramani indicated that the company will be exploring the automation of machine learning in forecasting and pricing, to make the service smarter and more efficient. Uber announced Professor Ghahramani’s appointment as Chief Scientist in March 2017, a position which involves strategic oversight of Uber’s AI Labs, the recently established machine learning and artificial intelligence research unit.

Further Information