Research

Projects: Currently (2019) all my research efforts go into Neural MCTS.

• Neural MCTS AlphaZero is learning how to solve combinatorial and strategic problems in games such as Chess. It makes sense to see if the AlphaZero approach can be extended to practically useful problems that we don't usually view as games. We currently focus on combinatorial problems solved by students in algorithms courses. A continuation of our scientific community games ideas but now using machines to do the playing instead of humans.
• Demeter Project / Aspect-Oriented Programming / Adaptive Programming (AP, since 1985)
  • Recent Projects:
    • AP-F (Functional Adaptive Programming, Bryan Chadwick)
      • DemeterF (AP-F for Java and C#)
      • apf-lib (AP-F for Scheme)
      • When AOP needs clever weaving algorithms: Illustration with program optimization (Ahmed Abdelmeged)
• Evergreen Project (since 1975, restarted in 2006)
  • Side-Choosing Game = Scientific Community Game = Specker Challenge Game (SCG)
  • Partial Satisfaction and Local versus Global Satisfaction. The most recent publication is in "Elemente der Mathematik" 2012 with Ernst Specker, published by the European Mathematical Society through Birkhaeuser.

Karl Lieberherr, Professor, CCIS.

Professor Lieberherr's research focuses on two areas: applied programming languages/software engineering, and the Scientific Community Game (SCG) to support innovation and learning through competition and collaboration. Applications of SCG are: (1) creating scientific discourse interactions between students through second order learning in traditional or MOOC courses. (2) creating innovations in STEM (Science, Technology, Engineering, Mathematics) areas through developing problem solving skills (3) learning to reverse engineer the skills of others through indirect questioning. (4) A cyberinfrastructure for teachers and researchers to program the "global brain".

He jointed Northeastern University in 1985 and established the Demeter research group, which developed Adaptive Programming (AP). AP abstracts out traversals from programs, which makes programs both simpler and more powerful. AP helps to control tangling between structure and behavior, and prevents unnecessary duplication of structural information. Aspect-oriented programming (AOP), a generalization of AP that Professor Lieberherr's group developed with Xerox PARC, advises arbitrary code, not just pure traversal code. DemeterF is the latest incarnation of AP using functional programming. DemeterF creates a balance between programming flexibility, traversal adaptability, and safety. DemeterF can be compiled with very minimal user intervention to multi-core architectures for fast execution. While the Demeter project initially focused on good design principles (such as the Law of Demeter) it also takes on the challenge of fast performance of adaptive programs on modern architectures.