Bryan Wayne Singer

bsinger+@cs.cmu.edu

Objective

Education

• Carnegie Mellon University
  
  • Ph.D. in Computer Science, expected December 2001
  • M.S. in Computer Science, August 2000
  • National Science Foundation (NSF) Graduate Fellowship, 1997-2000
  • Dissertation: Automating the Modeling and Optimization of the Performance of Signal Processing Algorithms
  • Research Advisor: Manuela Veloso
  • Research Areas: Machine learning and signal processing optimization
• Purdue University
  
  • B.S. in Computer Engineering with Highest Distinction and a Mathematics minor, May 1997
    
  • GPA: 4.0
    

Experience

• SPIRAL Research Group, Carnegie Mellon University
  Research Member, Fall 1998 - present
  • Developed and implemented the search engine in the SPIRAL system to produce platform adapted, optimized signal transform implementations
  • Collaborated with researchers from several universities in the development of the SPIRAL system
  • Designed methods for modeling performance of signal transform implementations using machine learning techniques by developing feature sets to describe such implementations
  • Developed a method for automatically constructing the fastest implementations of a signal transform by using learned models of performance
  • Employed a variety of machine learning techniques including decision and regression trees, neural networks, rule inducers, genetic algorithms, and reinforcement learning
• Computer Science Department, Carnegie Mellon University
  Teaching Assistant, Spring 1999, 2000
  • TA for undergraduate Artificial Intelligence course
• Center for Computing Sciences, Institute for Defense Analyses
  Intern, Summer 1998, 1999
  
  • Designed, improved, and evaluated algorithms for static program analysis of C code, including pointer analysis and interprocedural control and data flow analysis
  • Required TOP SECRET/SCI clearance
• National Security Agency (NSA)
  Director's Summer Program Participant, Summer 1997
  
  • Cryptological mathematics program
  • Improved a mathematical algorithm applying combinatorics, probability, and statistics to successfully handle a wider class of problems
  • Collaborated with team members and presented results to the Director of the NSA
  • Required TOP SECRET/SCI clearance
• Computer Science Department, University of Iowa
  Research Experience for Undergraduates (REU) Participant, Summer 1996
  
  • Extended the Probabilistic Theory Revision algorithm from revising propositional domain theories to revising relational domain theories
  • Created and tested a new inductive learning algorithm for relational theories
• Navistar International Technical Center
  Intern, Summer 1994, 1995
  
  • Enhanced existing engineering databases
  • Wrote UNIX shell scripts to perform verifications
  • Designed and organized a technical information center

Publications

Refereed

• Bryan Singer and Manuela Veloso. Stochastic search for signal processing algorithm optimization. In Proceedings of SC2001, 2001. To appear.
• Bryan Singer and Manuela Veloso. Learning to generate fast signal processing implementations. In Proceedings of the Eighteenth International Conference on Machine Learning, pages 529-536, 2001.
• Markus Pueschel, Bryan Singer, Manuela Veloso, and Jose Moura. Fast automatic generation of DSP algorithms. In Proceedings of the International Conference on Computational Science, Lecture Notes in Computer Science 2073, pages 97-106, 2001.
• Bryan Singer and Manuela Veloso. Learning to predict performance from formula modeling and training data. In Proceedings of the Seventeenth International Conference on Machine Learning, pages 887-894, 2000.
• Bryan Singer and Manuela Veloso. Learning state features from policies to bias exploration in reinforcement learning. In Proceedings of the Sixteenth National Conference on Artificial Intelligence, page 981, 1999.

Unrefereed

• Bryan Singer and Manuela Veloso. Stochastic search for signal processing algorithm optimization. Technical Report CMU-CS-01-137, Computer Science Department, Carnegie Mellon University, 2001.
• Bryan Singer and Manuela Veloso. Automated formula generation and performance learning for the FFT. Technical Report CMU-CS-00-123, Computer Science Department, Carnegie Mellon University, 2000.
• Bryan Singer and Manuela Veloso. Learning state features from policies to bias exploration in reinforcement learning. Technical Report CMU-CS-99-122, Computer Science Department, Carnegie Mellon University, 1999.