Amir Atiya

amiratiya(AT)link.net

amir(AT)alumni.caltech.edu

Professor
Computer Engineering Department, Cairo University

Ph.D. California Institute of Technology

Was born in Cairo, Egypt. He received his B.S. degree in 1982 from Cairo University (Egypt), and the M.S. and Ph.D. degrees in 1986 and 1991 from Caltech, Pasadena, CA, all in electrical engineering. He held positions in academia, as well as several positions in financial firms. From 1997 to 2001 he was a Visiting Associate at Caltech. He recently held research positions in the firms Simplex Risk Management, Hong Kong, Countrywide Corporation in Los Angeles, and Dunn Capital Markets, Florida.

He has been active in research in several fields. He received the Egyptian State Prize for Best Research in Science and Engineering, in 1994. He also received the Young Investigator Award from the International Neural Network Society in 1996. Recently, he received the prestigeous Kuwait Prize by the Kuwait Foundation for the Advancement of Sciences Currently, he is an Associate Editor for IEEE Transactions Neural Networks. He was guest editor of the special issue (July 2001) of IEEE Transactions Neural Networks on Neural Networks in Financial Engineering., and was guest editor of the special issue (September 2005) of IEEE Transactions Neural Networks on Adaptive Learning Systems in Communications Networks. He served as Program Co-chair of the conference IEEE Conference on Computational Intelligence in Financial Engineering (CIFER'03), Hong Kong, March 2003.

Publications

Research Interests


  • Statistical learning theory.
  • Neural networks.
  • Computational finance.
  • Trading system design.
  • Optimization theory.
  • Pattern recognition.
  • Time series forecasting.
  • Data compression.
  • Stochastic processes.
  • Monte Carlo methods.
  • Application of stochastic methods and optimization theory to communications networks.

    Most Recent Work:


    Analyzing the Maximum Drawdown:


    Theoretical analysis of the maximum drawdown risk measure, developing formulas relating it with the Sharpe Ratio, and analyzing its asymptotic behavior. See the papers maximum drawdown 1 (Journal of Applied Probability) and maximum drawdown 2 (Risk Magazine).

    A Large Scale Comparison of Machine Learning Models:


    A large scale and thorough study for 8 major machine learnine models for time series forecasting involving around 1000 time series from the M3 benchmark data. See the paper comparison 1 (to appear in Econometric Reviews journal).
    Based on this study, we participated in the NN3 Artificial Neural Networks & Computational Intelligence Forecasting Competition , which is a major international competition. Our rank was 6 out of 26 in one time series category and 5 out of 44 in other category. The model we used is based on neural networks and Gaussian process regression. See Results , where our model details can be downloaded.