HMAX image processing pipeline with coupled oscillator acceleration

In this paper we report on the performance of a coupled oscillator based implementation of the HMAX image-processing pipeline. Within this pipeline we have used coupled oscillator arrays to replace traditional Boolean logic with a Degree-of-Match (DoM) function that measures the L2 distance squared between two vectors in an n-dimensional space. We show that this operation can be used in three stages of the pipeline: 1) as a substitute for convolution in filtering operations, 2) as a computational kernel for pattern matching, and 3) as a distance function in a nearest neighbor classification algorithm. In this study, we have modeled the performance of the latter two and report our recognition results over a test set from the Neo Vision2 image database.

[1]  Balth. van der Pol,et al.  VII. Forced oscillations in a circuit with non-linear resistance. (Reception with reactive triode) , 1927 .

[2]  Tadashi Shibata,et al.  Coupled-Oscillator Associative Memory Array Operation , 2013 .

[3]  Frank C. Hoppensteadt,et al.  Synchronization of MEMS resonators and mechanical neurocomputing , 2001 .

[4]  John Carpenter,et al.  Image Processing Pipeline Based on Coupled Oscillator Models , 2014 .

[5]  Narayanan Vijaykrishnan,et al.  Saliency-driven dynamic configuration of HMAX for energy-efficient multi-object recognition , 2013, 2013 IEEE Computer Society Annual Symposium on VLSI (ISVLSI).

[6]  W. Porod,et al.  Spin torque oscillator models for applications in associative memories , 2012, 2012 13th International Workshop on Cellular Nanoscale Networks and their Applications.

[7]  David G. Lowe,et al.  Multiclass Object Recognition with Sparse, Localized Features , 2006, 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).

[8]  Renyuan Zhang,et al.  CMOS supporting circuitries for nano-oscillator-based associative memories , 2012, 2012 13th International Workshop on Cellular Nanoscale Networks and their Applications.

[9]  Tinku Acharya,et al.  Image Processing: Principles and Applications , 2005, J. Electronic Imaging.

[10]  J. Katine,et al.  Mutual phase-locking of microwave spin torque nano-oscillators , 2005, Nature.

[11]  S. Bhave,et al.  The resonant body transistor. , 2010, Nano letters.

[12]  Narayanan Vijaykrishnan,et al.  A hardware architecture for accelerating neuromorphic vision algorithms , 2011, 2011 IEEE Workshop on Signal Processing Systems (SiPS).

[13]  Narayanan Vijaykrishnan,et al.  Computational Architectures Based on Coupled Oscillators , 2014, 2014 IEEE Computer Society Annual Symposium on VLSI.

[14]  Y. Kuramoto,et al.  Dephasing and bursting in coupled neural oscillators. , 1995, Physical review letters.

[15]  F. Corinto,et al.  Synchronization in cellular spin torque oscillator arrays , 2012, 2012 13th International Workshop on Cellular Nanoscale Networks and their Applications.

[16]  Wolfgang Porod,et al.  Computational Study of Spin-Torque Oscillator Interactions for Non-Boolean Computing Applications , 2013, IEEE Transactions on Magnetics.

[17]  Thomas Serre,et al.  Object recognition with features inspired by visual cortex , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[18]  Narayanan Vijaykrishnan,et al.  Emulating Mammalian Vision on Reconfigurable Hardware , 2012, 2012 IEEE 20th International Symposium on Field-Programmable Custom Computing Machines.

[19]  Steven P. Levitan,et al.  Associative processing with coupled oscillators , 2013, ISLPED '13.

[20]  Saeed Karimimehr.  A novel face recognition system inspired by computational neuroscience , 2013, Eurocon 2013.

[21]  Christiaan Huygens.  Horologium oscillatorium 1673 , 1966 .

[22]  Vijaykrishnan Narayanan,et al.  Synchronized charge oscillations in correlated electron systems , 2014, Scientific Reports.

[23]  Yan Fang,et al.  Non-Boolean associative architectures based on nano-oscillators , 2012, 2012 13th International Workshop on Cellular Nanoscale Networks and their Applications.