B. Sc. Engineering in Computer Science and Engineering
Bangladesh University of Engineering and Technology
Contact details for Mr Swakkhar Shatabda
Thesis
Learning based Search and Optimization for Protein Structure Prediction Problem
Description
Protein structure prediction(PSP) is a central problem in Computational Biology. The problem of finding the stable and native conformation from an astronomically large number of possible conformations of a given amino acid sequence of a protein can be formulated as a search and optimisation problem. As a result recent years have seen some progress in applying search based optimisation technologies such as local search and constraint programming in protein structure prediction problem. However, enormous effort is required to achieve significant progress both in efficiency of the solvers and the quality of solutions. In my PhD work, I would like to explore the potential of advanced heuristics and learning techniques in the setting of local search for protein structure prediction problem.
Throughout my phd research, I would like to investigate the nature whether or not problems such as PSP exhibit the properties of structural repetitions of certain sub-problems.The structures could occur again and again inside a given problem or even across a given range of problems. Knowledge about such regular structures may be exploited to speed up the search process if the solutions of the sub-problems are known. I would develop new learning and exploitation techniques for learning about such regular structures with a view to improving the solution speed of the solver and the quality of the solution produced. Another important direction is to learn about the irregular structures in order to prune out the number of choices at the decision points or to decide the restarts at stagnation points of a search process. In my PhD, I intend to develop techniques that effectively learn and exploit such search control knowledge. Besides, reinforcement learning could be useful in utilising knowledge about past failures within the problem.
Supervisors
Professor Abdul Sattar, IIIS, Griffith University
Dr. Muhammad Abdul Hakim Newton, NICTA
Research expertise
- Bioinformatics
- Algorithms
- Search
- Optimization
Publications
- Atif Rahman, Swakkhar Shatabda, and Masud Hasan. An Approximation Algorithm for Sorting by Reversals and Transpositions. Journal of Discrete Algorithms, 6(3):449–457, Elsevier, 2008
- Swakkhar Shatabda, Atif Rahman, and Masud Hasan. An Approximation Algorithm for Bounded Degree Closest Phylogenetic 2nd Root Problem. In Proc. IAENG International Conference on Bioinformatics (ICB 2008), pages 157–161, Hong Kong, 2008
- Atif Rahman, Swakkhar Shatabda, and Masud Hasan. An Approximation Algorithm for Sorting by Reversals and Transpositions. In Proc.Workshop on Algorithms and Computation (WALCOM 2007), pages 97–108, Bangladesh, 2007