UAI 2015 - Invited Speakers

UAI 2015 is pleased to announce the following invited speakers:

---

Banquet Speaker

Raphael Slawinski

Mount Royal University

Should we go for it? Risk and decision-making in the mountains

One of the key aspects of alpinism is decision making. Climbers attempting a major ascent, be it in the Rockies or the Himalaya, face many decisions, ranging from the timing of the climb to the tactics used on it. However, probably the most difficult decision is whether to climb at all: drawing the line between acceptable and unacceptable risk. This decision is usually made with highly incomplete information, yet the potential consequences are severe. By drawing on examples of good - and bad - choices I have made in the mountains, I will discuss the interplay of intuition and rationality involved in deciding whether to go up or down.

Biographical details

Raphael Slawinski is a climbing physicist. During the week he teaches classical mechanics, electromagnetism or quantum mechanics at Mount Royal University in Calgary, Alberta, Canada, but he spends his weekends adventuring in the nearby Canadian Rockies, where he has many first ascents of difficult mixed routes. He has also been on expeditions to greater mountain ranges, such as Alaska, the Himalaya and the Karakoram. Here, in 2013, with Ian Welsted, he succeeded in the first ascent, alpine style, of K6 West (7040m), which was recognized with a Piolet d’Or award

Raphael Slawinski was drawn to Physics because of the insights it offered into the universe around us. Thus, while earning a major in Physics, he also completed a minor in Astrophysics. He went on to earn a master’s degree in Astrophysics, developing statistical methods to analyze x-ray telescope data, to try to understand the nature of gamma-ray bursts. He received his PhD degree in Geophysics, from the University of Calgary, in 1999, with computer models of how seismic waves propagate through fractured rock.

These days Dr. Slawinski is studying atmospheric physics. Some questions he asks himself are: Under what conditions could the Earth suffer a runaway greenhouse effect? What might be the effects of global warming on the seasonal cycle?

---

Keynote speaker

Peter Buhlmann

Swiss Federal Institute of Technology (ETH) Zurich

High-dimensional causal inference: exploiting the power of heterogeneous data

Understanding cause-effect relationships between variables is of great interest in many fields of science. Thus, inferring causal effects from data is a highly desirable but very ambitious goal. A major challenge is that many algorithms exhibit poor performance, particularly in the high-dimensional context where the number of variables can be much larger than sample size.

We present a novel methodology based on an invariance principle. It exploits the advantage of heterogeneity in larger datasets, arising from different experimental conditions (i.e. an aspect of "Big Data"). Despite fundamental identifiability issues, the method can be equipped with statistical confidence statements leading to more reliable results than alternative procedures based on graphical modeling. We finally discuss an application with large-scale gene knock-down experiments in yeast where computational and statistical methods have an interesting potential for prediction and prioritization of new experimental interventions.

Biographical details

Peter Bühlmann is Professor of Statistics at ETH Zürich. His research interests are in causal and high-dimensional statistical inference, machine learning and applications in the life sciences. He is a Fellow of the Institute of Mathematical Statistics and an elected Member of the International Statistical Institute. He presented a Medallion Lecture at the Joint Statistical Meetings 2009, a read paper to the Royal Statistical Society in 2010, and various named lectures. He wasEditor of the Annals of Statistics from 2010-2012.

---

Keynote speaker

David MacKay FRS

Cambridge University

Why climate change action is difficult, and how we can make a difference

I will discuss several reasons why climate-change action is difficult.
First, the climate has a long-lasting response to cumulative emissions of carbon. This inconvenient truth implies that it is not enough to cut the emissions rate by some fraction such as 50% or 80%. Climate change will stop increasing only when the net emissions rate is cut to zero; and, equally inconveniently, undoing climate change requires negative emissions.
Second, the public and many decision makers have been misled by myths and wishful thinking about the scale of action required to decarbonize the energy system.
Third, effective climate change action will require the large-scale deployment of low-carbon technologies, most of which are expensive.

We can make a difference by
a) getting involved in innovation and research and development of lower-cost solutions;
b) supporting a numerate approach to energy policy; and
c) supporting the development of open-source energy models for all countries.

Biographical details

David MacKay FRS is the Regius Professor of Engineering at the University of Cambridge. After completing his PhD in Computation and Neural Systems at the California Institute of Technology in 1991, he worked at the Cavendish Laboratory in Cambridge first as a research fellow and then as a member of the teaching faculty. From 1988 to 2008 his research interests included reliable computation with unreliable hardware; error-correcting codes for reliable communication; Bayesian methods for machine learning; and information-theory-based communication systems to allow people with disabilities to communicate efficiently. His best-selling textbook,Information Theory, Inference, and Learning Algorithms (free online here), was published in 2003. In 2008, to try to improve the quality of the public debate about energy options, he wrote the critically acclaimed book 'Sustainable Energy - without the hot air' (free online here). From 2009 to 2014 he served the UK Government as Chief Scientific Advisor to the Department of Energy and Climate Change (DECC). Notable open-source projects he was involved with at DECC include the UK’s “2050 Calculator” (here) and the "Global Calculator” (here).

---

Keynote speaker

David Silver

Google DeepMind

Deep Reinforcement Learning

In this talk I will discuss how reinforcement learning (RL) can be combined with deep learning (DL). There are several ways to combine DL and RL together, including value-based, policy-based, and model-based approaches with planning. Several of these approaches have well-known divergence issues, and I will present simple methods for addressing these instabilities. These methods have achieved notable success in the Atari 2600 domain. I will present recent a selection of recent results that improve on the published state-of-the-art in Atari and other challenging domains. Finally, I will discuss how RL can be used to improve DL, even when the native problem is supervised or unsupervised learning.

Biographical details

David's research focuses on reinforcement learning, planning and control. David leads a research group on reinforcement learning at Google DeepMind. His recent work has focused on combining reinforcement learning with deep learning, including a program that learns to play Atari games directly from pixels (Nature 2015). David holds a Royal Society University Research Fellowship and is a lecturer at University College London. His PhD (supervised by Rich Sutton at the University of Alberta) was on reinforcement learning in Computer Go, which co-introduced the algorithms used in the first master level Go programs. In his previous life, he was CTO at Elixir Studios and Lead Programmer on Republic: the Revolution.

---