Combining Software and Hardware LCS for Lightweight On-Chip Learning

In this paper we present a novel two-stage method to realize a lightweight but very capable hardware implementation of a Learning Classifier System for on-chip learning. Learning Classifier Systems (LCS) allow taking good run-time decisions, but current hardware implementations are either large or have limited learning capabilities. In this work, we combine the capabilities of a software-based LCS, the XCS, with a lightweight hardware implementation, the LCT, retaining the benefits of both. We compare our method with other LCS implementations using the multiplexer problem and evaluate it with two chiprelated problems, run-time task allocation and SoC component parameterization. In all three problem sets, we find that the learning and selfadaptation capabilities are comparable to a full-fledged system, but with the added benefits of a lightweight hardware implementation, namely small area size and quick response time. Given our work, autonomous chips based on Learning Classifier Systems become feasible.

[1]  Coniferous softwood.  GENERAL TERMS , 2003 .

[2]  Wolfgang Rosenstiel,et al.  Generic Self-Adaptation to Reduce Design Effort for System-on-Chip , 2009, 2009 Third IEEE International Conference on Self-Adaptive and Self-Organizing Systems.

[3]  Pier Luca Lanzi,et al.  Mining interesting patterns from hardware-software codesign data with the learning classifier system XCS , 2003, The 2003 Congress on Evolutionary Computation, 2003. CEC '03..

[4]  Pier Luca Lanzi,et al.  Evolving classifiers on field programmable gate arrays: Migrating XCS to FPGAs , 2006, J. Syst. Archit..

[5]  Walter Stechele,et al.  Learning Classifier Tables for Autonomic Systems on Chip , 2008, GI Jahrestagung.

[6]  Fernando Morgado Dias,et al.  Artificial neural networks: a review of commercial hardware , 2004, Eng. Appl. Artif. Intell..

[7]  Wolfgang Rosenstiel,et al.  Towards a Framework and a Design Methodology for Autonomic SoC , 2005, Second International Conference on Autonomic Computing (ICAC'05).

[8]  Stewart W. Wilson.  Generalization in the XCS Classifier System , 1998 .

[9]  Bernard Widrow,et al.  Neural networks: applications in industry, business and science , 1994, CACM.

[10]  Karl Goser,et al.  On the degradation of p-MOSFETs in analog and RF circuits under inhomogeneous negative bias temperature stress , 2003, 2003 IEEE International Reliability Physics Symposium Proceedings, 2003. 41st Annual..

[11]  Marco Dorigo,et al.  Alecsys and the AutonoMouse: Learning to control a real robot by distributed classifier systems , 2004, Machine Learning.

[12]  Stewart W. Wilson.  Classifier systems and the animat problem , 2004, Machine Learning.

[13]  E. Nowak,et al.  High-performance CMOS variability in the 65-nm regime and beyond. IBM J Res And Dev , 2006 .

[14]  Stewart W. Wilson.  ZCS: A Zeroth Level Classifier System , 1994, Evolutionary Computation.

[15]  Martin V. Butz,et al.  Analysis and Improvement of Fitness Exploitation in XCS: Bounding Models, Tournament Selection, and Bilateral Accuracy , 2003, Evolutionary Computation.

[16]  H. Crichton-Miller.  Adaptation , 1926 .

[17]  Narayanan Vijaykrishnan,et al.  A Hardware Efficient Support Vector Machine Architecture for FPGA , 2008, 2008 16th International Symposium on Field-Programmable Custom Computing Machines.

[18]  Wolfgang Rosenstiel,et al.  Evaluation of the Learning Classifier System XCS for SoC run-time control , 2008, GI Jahrestagung.

[19]  Hartmut Schmeck,et al.  Organic Control of Traffic Lights , 2008, ATC.

[20]  Stewart W. Wilson.  Classifier Fitness Based on Accuracy , 1995, Evolutionary Computation.

[21]  Martin V. Butz,et al.  An algorithmic description of XCS , 2000, Soft Comput..

[22]  Narayanan Vijaykrishnan,et al.  Reliability concerns in embedded system designs , 2006, Computer.