Paul Werbos, National Science Foundation
Dr. Werbos has core responsibility for the Adaptive and Intelligent Systems (AIS) area within the Controls, Networks and Computational Intelligence (CNCI) Program of ECS. He is also leading the development of a new Cybersystems thrust within the Integrative Hybrid and Complex Systems (IHCS) program of ECS. He is the ECS representative for the CLEANER initiative, for biocomplexity (MUSES), and for Collaborative Research in Computational NeuroScience. He is one of the two ECS representatives for cyberinfrastructure. He has special interest in efforts to exploit higher levels of true computational intelligence in these areas, and in efforts which can seriously increase the probability that we achieve global sustainability. In 1994, he initiated an SBIR topic on fuel cell and electric cars which he coordinated for several years. He was part of the group which proposed and led NSF's earlier initiative in Learning and Intelligent Systems, and assisted the follow-on in Information Technology Research. He has at times handled the ECS areas in electric power and wireless communications when there were gaps in those areas.
Dr. Werbos is an elected member of the Administrative Committee (AdCom) of the IEEE Computational Intelligence Society, which he represents on the IEEE-USA Energy Policy Committee. (See www.ieeeusa.org/policy/energy_strategy.ppt.) He also serves on the AdCom of the IEEE Industrial Electronics Society, and the Governing Board of the International Neural Network Society (INNS). He was one of the three original two-year Presidents of INNS. He is a Fellow of the IEEE, and has won its Neural Network Pioneer Award, for the discovery of the backpropagation algorithm and other basic neural network learning designs. He also serves on the Planning Committee of the ACUNU Millennium Project (see www.stateofthefuture.org), whose annual report on the future tends to lead global lists of respected reports on the long-term future. In 2002, he and John Mankins of NASA initiated and ran the NASA-NSF-EPRI initiative on enabling technologies for space solar power. In 2003, he participated on the interagency working group for the Climate Change Technology Program. At the 2005 Space Development Conference in Arlington, he was invited to present a new strategy for sustainable exploration and development of space, drawing in part on previous work funded by NSF.
In addition to his core interests at NSF, Dr. Werbos has interest in larger questions relating to consciousness, the foundations of physics, and human potential. His 1974 Harvard Ph.D. thesis has been reprinted in its entirety, along with related papers, in his book The Roots of Backpropagation: From Ordered Derivatives to Neural Networks and Political Forecasting, Wiley, 1994. Some of work on high performance computing is described in P. Werbos, Backwards differentiation in AD and Neural Nets: Past Links and New Opportunities. In Martin Bucker, George Corliss, Paul Hovland, Uwe Naumann & Boyana Norris (eds), Automatic Differentiation: Applications, Theory and Implementations, Springer (LNCS), New York, 2005.
Prior to arriving full-time at NSF in 1989, Dr. Werbos worked since 1979 at the Energy Information Administration (EIA) of the Department of Energy. He initially worked in the evaluation of energy models, forecasts and analyses; this required spanning the gamut from decoding undocumented FORTRAN to evaluating the implications for the future of humanity. He later became lead analyst for long-term energy futures, and developed the econometric models used in EIA's Annual Energy Outlook for industrial and transportation energy demand and for oil and gas production. He served on Carter's Global 2000 Phase II interagency task force. His model of industrial energy demand played a major role in the Stanford Energy Modeling Forum study of industrial demand, and resulted in several papers, including two in Energy: The International Journal, March/April 1990. Before that he spent a year as an IPA at the Census Use Research center as a mathematical statistician, and taught for 3 and a half years at the University of Maryland in the public policy area. Before teaching, he spent two years at the MIT Cambridge Project adding new capabilities for data mining and modeling to a user-oriented software package written in FORTRAN and PL/1 for the Multics operating system.
He holds four degrees from Harvard and the London School of Economics in: (1) economics; (2) international political systems, emphasizing European economic institutions; (3) applied mathematics, with a major in quantum physics and a minor in decision and control; (4) applied mathematics for an interdisciplinary PhD. Prior to that, during high school, he obtained an FCC First Class Commercial Radiotelephone license, and took undergraduate and graduate mathematics courses at Princeton and the University of Pennsylvania.