TURINGEVAL - Key Persons

Jakob Shimer is from Lake Charles, Louisiana. He received his BA from Reed College in Mathematics and Computer Science. His interests include Artificial Intelligence and Machine Learning. Jakob specializes in the development and implementation of machine learning methods for learning and predicting probability distributions.

Nick Candau is from San Francisco. He studied Cell Biology at San Francisco State University, Intelligence Management at Henley-Putnam, and Political Science at University of California, Berkeley. In 2016 he was invited to participate in IARPA's Hybrid Forecast Challenge. In addition to handling the day-to-day operations, Nick specializes in logic, knowledge engineering, and business development strategy and has worked with the team to further those areas.

Peter Cheeseman received his B. Sc. (Hons) from Melbourne University (Australia) in 1972, majoring in Physics and Applied Mathematics; his M. Phil. in 1974 from Waikato University (New Zealand), majoring in Applied Mathematics, and his Ph.D. in 1978 from Monash University (Australia) in artificial intelligence (AI). After 4 years spent teaching advanced computing (University of Technology Sydney), he moved to SRI International (California) where he continued his AI research in planning, search and reasoning under uncertainty. At SRI he developed new algorithms for efficient Maximum Entropy inference that foreshadowed later Bayes Net algorithms, now standard in AI. In 1985 he co-founded the annual Uncertainty in AI conference. While at SRI, he developed a new approach to reasoning with spatial uncertainty called SLAM (Simultaneous Localization And Mapping). This SLAM approach is now standard for robot navigation and tolerance stacking. In 1985, he moved to NASA Ames Research Center (Bay Area) where he extended his research on reasoning under uncertainty to automatic discovery of structure in data, and unsupervised learning (i.e., automated discovery of natural classes in data), leading to the influential, and widely used AutoClass program. During this period he also discovered the foundational result that computationally hard problems occur at the solvability phase transition boundary. In 1991 he received US patent number 5,302,509 for a new method of DNA sequencing he invented. He and his collaborators were the first to apply Bayesian learning methods to the problem of 2-D super-resolution from multiple images and more recently extended these methods to produce super-resolved 3-D surfaces from multiple view-point images. In 2004 he moved back to Sydney Australia to head the Symbolic Machine Learning and Knowledge Acquisition group at NICTA. His team's research at NICTA worked on developing a general integrated AI system. In 2006, he moved back to the Bay Area to focus on the development of a new efficient internal combustion engine. In 2007 he was awarded the AAAI classic paper prize for his AutoClass paper.