AI/ML

KEDGE pursues core research in AI/ML, especially in Constraints and AVs

Summary

     Kedge Systems has a good experience in doing research with high theoretical and industrial impact. Recently, our focus is in building intelligent AI/ML systems (with applications in constraint systems and autonomous vehicles).

     As an example, one of our past works, NiVER SAT Preprocessor, via its deriviatives like SatElite preprocessor etc.,, is in use in all modern industrial SAT solvers (for wide range of applications, including Hardware/Software verification, artificial intelligence, automated theorem proving). Our past work in using (Tree) Decomposition techniques for solving constraint systems (Tree-of-BDDs) also had a huge industrial and academic impact.

      Recent Research Reports

• S. Subbarayan, B. Kathatharan, "Machine Learning for Predicting Constraints: An Empirical Study", Kedge Technical Report No. 2022-002, 09-Nov-2022 (PDF)

• S. Subbarayan, B. Kathatharan, "Machine Learning for Predicting Constraints", Kedge Technical Report No. 2022-001, 05-Aug-2022 (PDF)
  

      Related US/EU Patents

• PRODUCT CONFIGURATION (using DAOG), Filed US 2015/0331974 A1 (16-May-2014, with Henrik Reif Andersen, for Configit A/S)

• PRODUCT CONFIGURATION (using JIT Tree-of-BDDs), Filed US 2015/0120490 A1 (31-Oct-2013, with Henrik Reif Andersen, Henrik Hulgaard, for Configit A/S)

      Past Research (AI, SAT, Genetic Algorithms)

• S. Subbarayan, "On Exploiting Structures for Constraint Solving", PhD Thesis, IT-University of Copenhagen, Denmark, 196 pages. April 2008, ISBN : 978-87-7949-173-1

• S. Subbarayan, "Efficient Reasoning for Nogoods in Constraint Solvers with BDDs", Springer LNCS Proceedings of the PADL symposium, Jan-2008, San Francisco, USA

• S. Subbarayan, "An Empirical Comparison of CSP Decomposition Methods", Proceedings of the Doctoral Program, CP conference, Sept-2007, Providence, USA.

• S. Subbarayan, L. Bordeaux and Y Hamadi, "Knowledge Compilation Properties of Tree-of-BDDs", Proceedings of the AAAI conference, July-2007, Vancouver, Canada

• S. Subbarayan, and , H R Andersen, "Backtracking Procedures for Hypertree, HyperSpread and Connected Hypertree Decomposition of CSPs", Proceedings of the IJCAI conference, Jan-2007, Hyderabad, India

• S. Subbarayan, L. Bordeaux and Y Hamadi, "On Hybrid SAT Solving using Tree Decompositions and BDDs", Technical Report MSR-TR-2006-28, Microsoft Research, Cambridge, UK (2006)

• S. Subbarayan, and H R Andersen, "Linear Functions for Interactive Configuration Using Join Matching and CSP Tree Decomposition", Proceedings of the IJCAI-2005 Configuration Workshop, July- 2005, Edinburgh, UK

• S. Subbarayan, and H R Andersen, " Integrating a Variable Ordering Heuristic with BDDs and CSP Decomposition Techniques for Interactive Configurators", Research Note, April- 2005

• S. Subbarayan, "Integrating CSP Decomposition Techniques and BDDs for Compiling Configuration Problems", Proceedings of the International Conference on Integration of AI and OR Techniques in Constraint Programming for Combinatorial Optimization Problems, CP-AI-OR, (LNCS 3524) May-2005, Prague, Czech Republic

• S. Subbarayan, R M Jensen, T. Hadzic, H R Andersen, J Møller, and H Hulgaard, "Comparing Two Implementations of a Complete and Backtrack-Free Interactive Configurator", Proceedings of the CP-04 Workshop on CSP Techniques with Immediate Application, 97-111, Sept-2004, Toronto, Canada

• T. Hadzic, S. Subbarayan, R M Jensen, H R Andersen, J Møller, and H Hulgaard, "Fast Backtrack-Free Product Configuration Using a Precompiled Solution Space Representation", Proceedings of International Conference on Economic, Technical and Organisational aspects of Product Configuration Systems, June-2004 , Copenhagen

• S Subbarayan and DK Pradhan, “NiVER: Non Increasing Variable Elimination Resolution”, In Selected Revised Papers of International Conference on Theory and Applications of Satisfiability Testing Conference, SAT-2004, (Springer LNCS), May-2004, Vancouver, Canada

• G. Andal Jayalakshmi, S. Sathiamoorthy, and R. Rajaram –“A Hybrid Genetic Algorithm- A New Approach to solve Traveling Salesman Problem”, International Journal of Computational Engineering Science Vol.2 No.2, pp339–355, Imperial College Press. June 2001

• S. Sathiamoorthy –“LaySeq: A New Representation for Non-Slicing Floorplans”, Proceedings of 6th IEEE VLSI Design and Test Workshop, pp 317-327, Bangalore, India, August 29-31, 2002.

• S. Sathiamoorthy, and G. Andal Jayalakshmi – “Hybrid Genetic Algorithm for VLSI Macro Cell Layout”, in 6th On-line World Conference on Soft Computing in Industrial Applications, September 10-24, 2001. Published as a chapter in book titled “Soft Computing in Industry –Recent Applications”, Springer-Verlag. pp 797-804. One among WSC6 Best Papers.

• S. Sathiamoorthy, and G. Andal Jayalakshmi –“An Adaptive Hybrid Genetic algorithm for Macro Cell Layout”, Proceedings of International Symposium on Artificial Intelligence 2001.December 18-20, 2001

• S. Sathiamoorthy, G. Andal Jayalakshmi, and R. Rajaram –“Traveling Salesman Problem – A new approach using a Hybrid Genetic Agorithm”, Proceedings of National Conference on Advanced Computing 2001, pp 135-142. February 2-3, 2001