Real-time scheduling of mixture-of-experts systems with limited resources

Mixture-of-Experts (MoE) systems solve intricate problems by combining results generated independently by multiple computational models (the "experts"). Given an instance of a problem, the responsibility of an expert measures the degree to which the expert's output contributes to the final solution. Brain Machine Interfaces are examples of applications where an MoE system needs to run periodically and expert responsibilities can vary across execution cycles. When resources are insufficient to run all experts in every cycle, it becomes necessary to execute the most responsible experts within each cycle. The problem of adaptively scheduling experts with dynamic responsibilities can be formulated as a succession of optimization problems. Each of these problems can be solved by a known technique called "task compression" using explicit mappings described in this paper to relate expert responsibilities to task elasticities. A novel heuristic is proposed to enable real-time execution rate adaptation in MoE systems with insufficient resources. In any given cycle, the heuristic uses sensitivity analysis to test whether one of two pre-computed schedules is the optimal solution of the optimization problem to avoid re-optimization when the test result is positive. These two candidate schedules are the schedule used in the previous cycle and the schedule pre-computed by the heuristic during the previous cycle, using future responsibilities predicted by the heuristic's responsibility predictor. Our heuristic significantly reduces the scheduling delay in the execution of experts when re-execution of the task-compression algorithm is not needed from O(N2) time, where N denotes the number of experts, to O(N) time. Experimental evaluation of the heuristic on a test case in motor control shows that these time savings occur and scheduled experts' deadlines are met in up to 90% of all cycles. For the test scenario considered in the paper, the average output error of a real-time MoE system due to the use of limited resources is less than 7%.

[1]  Giuseppe Lipari,et al.  Elastic Scheduling for Flexible Workload Management , 2002, IEEE Trans. Computers.

[2]  Antonia Ghiselli,et al.  Very Large ensemble ocean forecasting experiment using the Grid computing infrastructure , 2008 .

[3]  Renato J. O. Figueiredo,et al.  A New Architecture for Deriving Dynamic Brain-Machine Interfaces , 2006, International Conference on Computational Science.

[4]  Michael D. Lemmon,et al.  Generalized Elastic Scheduling for Real-Time Tasks , 2009, IEEE Transactions on Computers.

[5]  Robert A. Jacobs,et al.  Hierarchical Mixtures of Experts and the EM Algorithm , 1993, Neural Computation.

[6]  Mokhtar S. Bazaraa,et al.  Nonlinear Programming: Theory and Algorithms , 1993 .

[7]  Daeyeol Lee,et al.  Behavioral Context and Coherent Oscillations in the Supplementary Motor Area , 2022 .

[8]  Simon Haykin,et al.  Neural Networks and Learning Machines , 2010 .

[9]  Thomas M. Hamill,et al.  Ensemble Data Assimilation with the NCEP Global Forecast System , 2008 .

[10]  Lei Xu,et al.  Application of Mixture of Experts Model To Financial Time Series Forecasting , 1995 .

[11]  Yanjun Qi,et al.  A Mixture of Experts Approach for Protein-Protein Interaction Prediction , 2005 .

[12]  Ziv Bar-Joseph,et al.  A mixture of feature experts approach for protein-protein interaction prediction , 2007, BMC Bioinformatics.

[13]  S. Caselli,et al.  A Technique for Adaptive Scheduling of Soft Real-Time Tasks , 2005, Real-Time Systems.

[14]  Renato J. O. Figueiredo,et al.  Towards Real-Time Distributed Signal Modeling for Brain-Machine Interfaces , 2007, International Conference on Computational Science.

[15]  Andreas S. Weigend,et al.  Time Series Prediction: Forecasting the Future and Understanding the Past , 1994 .

[16]  D M Wolpert,et al.  Multiple paired forward and inverse models for motor control , 1998, Neural Networks.

[17]  Ping Guo,et al.  Mixture of Experts for Stellar Data Classification , 2005, ISNN.

[18]  Vasant Honavar,et al.  Mixture of experts models to exploit global sequence similarity on biomolecular sequence labeling , 2009, BMC Bioinformatics.

[19]  R. J. Graham,et al.  Joint Medium-Range Ensembles from The Met. Office and ECMWF Systems , 2000 .

[20]  José Carlos Príncipe,et al.  Brain-Machine Interface Engineering , 2006, Brain-Machine Interface Engineering.