Hardware Accelerator for Probabilistic Inference in 65-nm CMOS

A hardware accelerator is presented to compute the probabilistic inference for a Bayesian network (BN) in distributed sensing applications. For energy efficiency, the accelerator is operated at a near-threshold voltage of 0.5 V, while achieving a maximum clock frequency of 33 MHz. Clique-tree message passing algorithm is leveraged to compute the probabilistic inference. The theoretical maximum size of a factor that the proposed hardware accelerator can handle is 2(8×20)=160 entries, which is sufficient for handling massive BNs, such as PATHFINDER, MUNIN, and so on (>1000 nodes). A Logical Alarm Reduction Mechanism (ALARM) BN is used to benchmark the performance of the accelerator. The accelerator consumes 76 nJ to execute the ALARM network using a clique-tree message-passing algorithm, while the same algorithm executed on an ultralow-power microcontroller consumes 20 mJ.

[1]  Radford M. Neal.  Pattern Recognition and Machine Learning , 2007, Technometrics.

[2]  David Blaauw,et al.  Circuits for a Cubic-Millimeter Energy-Autonomous Wireless Intraocular Pressure Monitor , 2011, IEEE Transactions on Circuits and Systems I: Regular Papers.

[3]  Yanqing Zhang,et al.  Body Sensor Networks: A Holistic Approach From Silicon to Users , 2012, Proceedings of the IEEE.

[4]  David Blaauw,et al.  A cubic-millimeter energy-autonomous wireless intraocular pressure monitor , 2011, 2011 IEEE International Solid-State Circuits Conference.

[5]  David Blaauw,et al.  Near-Threshold Computing: Reclaiming Moore's Law Through Energy Efficient Integrated Circuits , 2010, Proceedings of the IEEE.

[6]  Marina L. Gavrilova,et al.  Belief trees and networks for biometric applications , 2011, Soft Comput..

[7]  Chung-Hsien Wu,et al.  A Bayesian neural network chip design for speech recognition system , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[8]  Nir Friedman,et al.  Probabilistic Graphical Models: Principles and Techniques - Adaptive Computation and Machine Learning , 2009 .

[9]  Gregory F. Cooper,et al.  The ALARM Monitoring System: A Case Study with two Probabilistic Inference Techniques for Belief Networks , 1989, AIME.

[10]  Zhi-Hui Kong,et al.  Hardware realization of a medical diagnostic system based on Probabilistic CMOS (PCMOS) technology , 2008, 2008 IEEE International Symposium on VLSI Design, Automation and Test (VLSI-DAT).

[11]  Gerhard Goos,et al.  Ambient Intelligence , 2015, Lecture Notes in Computer Science.

[12]  Judea Pearl,et al.  Fusion, Propagation, and Structuring in Belief Networks , 1986, Artif. Intell..

[13]  G.E. Moore,et al.  Cramming More Components Onto Integrated Circuits , 1998, Proceedings of the IEEE.

[14]  S. Marra,et al.  FPGA implementation of Bayesian neural networks for a stand-alone predictor of pollutants concentration in the air , 2004, 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541).

[15]  John Wawrzynek,et al.  High-throughput bayesian computing machine with reconfigurable hardware , 2010, FPGA '10.

[16]  Bohn Stafleu van Loghum,et al.  Online … , 2002, LOG IN.

[17]  L. Baum,et al.  A Maximization Technique Occurring in the Statistical Analysis of Probabilistic Functions of Markov Chains , 1970 .

[18]  Pakpong Chirarattananon,et al.  Adaptive control of a millimeter-scale flapping-wing robot , 2014, Bioinspiration & biomimetics.

[19]  Lida Xu,et al.  The internet of things: a survey , 2014, Information Systems Frontiers.