Brain derived vision algorithm on high performance architectures. (English) Zbl 1192.68012
Summary: Even though computing systems have increased the number of transistors, the switching speed, and the number of processors, most programs exhibit limited speedup due to the serial dependencies of existing algorithms. Analysis of intrinsically parallel systems such as brain circuitry have led to the identification of novel architecture designs, and also new algorithms than can exploit the features of modern multiprocessor systems. In this article we describe the details of a Brain Derived Vision (BDV) algorithm that is derived from the anatomical structure, and physiological operating principles of thalamo-cortical brain circuits. We show that many characteristics of the BDV algorithm lend themselves to implementation on IBM CELL architecture, and yield impressive speedups that equal or exceed the performance of specialized solutions such as FPGAs. Mapping this algorithm to the IBM CELL is non-trivial, and we suggest various approaches to deal with parallelism, task granularity, communication, and memory locality. We also show that a cluster of three PS3s (or more) containing IBM CELL processors provides a promising platform for brain derived algorithms, exhibiting speedup of more than \(140 \times \) over a desktop PC implementation, and thus enabling real-time object recognition for robotic systems.
MSC:
| 68M07 | Mathematical problems of computer architecture |
| 68T45 | Machine vision and scene understanding |
| 68W10 | Parallel algorithms in computer science |
References:
| [1] | Carruthers, J., Hammerstrom, D., Colwell, B., Bourianoff, G., Zhirnov, V.: ITRS emerging research architecture working group white paper. http://www.itrs.net/Links/2005ITRS/Home2005.htm . Accessed April 2009 (2005) |
| [2] | NAE: Grand challenges for engineering. http://www.engineeringchallenges.org/ . Accessed April 2009 (2008) |
| [3] | Granger, R.: Brain circuit implementation: high-precision computations from low-precision components. In: Toward Replacement Parts for the Brain, pp. 277–294. MIT Press, Cambridge (2004) |
| [4] | Itti L., Koch C.: Computational modelling of visual attention. Nat. Rev. Neurosci. 2, 194–203 (2001) · doi:10.1038/35058500 |
| [5] | Granger R.: Engines of the brain: the computational instruction set of human cognition. AI Mag. 27, 15–32 (2006) |
| [6] | Furlong, J., Felch, A., Jayram, M.N., Dutt, N., Nicolau, A., Veidenbaum, A., Chandrashekar, A., Granger, R.: Novel brain-derived algorithms scale linearly with number of processing elements. In: Proceedings of ParaFPGA Conference: Parallel Computing with FPGAs (2007) · Zbl 1160.68686 |
| [7] | Felch, A., Jayram, M.N., Chandrashekar, A., Furlong, J., Dutt, N., Granger, R., Nicolau, A., Veidenbaum, A.: Accelerating brain circuit simulations of object recognition with cell processors. In: International Workshop on Innovative Architecture for Future Generation Processors and Systems, pp. 33–42 (2007) · Zbl 1160.68686 |
| [8] | Granger R., Hearn R.: Models of thalamocortical system. Scholarpedia 2(11), 1796 (2007) · doi:10.4249/scholarpedia.1796 |
| [9] | Izhikevich E.M., Edelman G.M.: Large-scale model of mammalian thalamocortical systems. Proc. Natl. Acad. Sci. 105, 3593–3598 (2008) · doi:10.1073/pnas.0712231105 |
| [10] | Serre T., Wolf L., Bileschi S., Riesenhuber M., Poggio T.: Robust object recognition with cortex-like mechanisms. IEEE Trans. Pattern Anal. Mach. Intell. 29, 411–426 (2007) · doi:10.1109/TPAMI.2007.56 |
| [11] | Krichmar J., Seth A., Nitz D., Fleischer J., Edelman G.: Spatial navigation and causal analysis in a brain-based device modeling cortical-hippocampal interactions. NeuroInformatics 3, 197–221 (2005) · doi:10.1385/NI:3:3:197 |
| [12] | Rodriguez A., Whitson J., Granger R.: Derivation and analysis of basic computational operations of thalamocortical circuits. J. Cogn. Neurosci. 16, 856–877 (2004) · doi:10.1162/089892904970690 |
| [13] | Lynch G., Granger R.: Big Brain: The Origins and Future of Human Intelligence, 1st edn. Palgrave Macmillan, New York (2008) |
| [14] | Amdahl, G.M.: Validity of the single processor approach to achieving large scale computing capabilities. In: AFIPS Conference Proceedings (1967) |
| [15] | Kahle J.A., Day M.N., Hofstee H.P., Johns C.R., Maeurer T.R., Shippy D.: Introduction to the CELL multiprocessor. IBM J. Res. Develop. 49, 589–604 (2005) · doi:10.1147/rd.494.0589 |
| [16] | Gschwind M., Hofstee H., Flachs B., Hopkins M., Watanabe Y., Yamazaki T.: Synergistic processing in cell’s multicore architecture. IEEE Micro 26, 10–24 (2006) · doi:10.1109/MM.2006.41 |
| [17] | Brain derived vision on IBM CELL: http://sourceforge.net/projects/bdv-cell . Accessed April 2009 (2007) |
| [18] | Mead C.: Analog VLSI and Neural Systems, 1st edn. Addison Wesley Publishing Company, USA (1989) · Zbl 0715.68002 |
| [19] | Olshausen B.A., Field D.J.: How close are we to understanding v1?. Neural Comput. 17(8), 1665–1699 (2005) · Zbl 1071.92009 · doi:10.1162/0899766054026639 |
| [20] | Hubel, D.H.: Eye, Brain, and Vision. W.H. Freeman, USA (1995) |
| [21] | Lee, H., Chaitanya, E., Ng, A.Y.: Sparse deep belief net model for visual area v2. In: Advances in Neural Information Processing Systems, vol. 20. MIT Press, Cambridge (2007) |
| [22] | Felch, A.C.: From synapse to psychology: emergence of a language, speech, and vision engine from bottom-up brain modeling. Ph.D. dissertation, University of California, Irvine (2006) |
| [23] | Foldiak P., Endres D.: Sparse coding. Scholarpedia 3(1), 2984 (2008) · doi:10.4249/scholarpedia.2984 |
| [24] | Marr D., Nishihara H.K.: Representation and recognition of the spatial organization of three-dimensional shapes. Proc. R. Soc. Lond. B 200, 269–294 (1978) · doi:10.1098/rspb.1978.0020 |
| [25] | Coultrip R., Granger R., Lynch G.: A cortical model of winner-take-all competition via lateral inhibition. Neural Netw. 5, 47–54 (1992) · doi:10.1016/S0893-6080(05)80006-1 |
| [26] | Stringer S.M., Rolls E.T.: Invariant object recognition in the visual system with novel views of 3d objects. Neural Comput. 14(11), 2585–2596 (2002) · Zbl 1057.68676 · doi:10.1162/089976602760407982 |
| [27] | Riesenhuber M., Poggio T.: Hierarchical models of object recognition in cortex. Nature Neurosci. 2(11), 1019–1025 (1999) · doi:10.1038/14819 |
| [28] | George, D., Hawkins, J.: A hierarchical bayesian model of invariant pattern recognition in the visual cortex. In: IJCNN ’05: Proceedings of IEEE International Joint Conference on Neural Networks, vol. 3, pp. 1812–1817 (2005) |
| [29] | Oram, A., Wilson, G.: Beautiful Code: Leading Programmers Explain How They Think. O’Reilly Media, Inc., USA (2007) |
| [30] | IBM: CELL broadband engine programming handbook–version 1.1 (2005) |
| [31] | Eichenberger, A.E., O’Brien, K., O’Brien, K., Wu, P., Chen, T., Oden, P.H., Prener, D.A., Shepherd, J.C., So, B., Sura, Z., Wang, A., Zhang, T., Zhao, P., Gschwind, M.: Optimizing compiler for the cell processor. In: PACT ’05: Proceedings of the 14th International Conference on Parallel Architectures and Compilation Techniques, pp. 161–172. IEEE Computer Society, Washington, DC (2005) |
| [32] | Thorpe S., Fize D., Marlot C.: Speed of processing in the human visual system. Nature 381, 520–522 (1996) · doi:10.1038/381520a0 |
| [33] | Mattson T., Sanders B.A., Massingill B.L.: Patterns for Parallel Programming, 1st edn. Addison-Wesley, USA (2004) · Zbl 1002.68586 |
| [34] | Lee, E.A., Parks, T.M.: Dataflow Process Networks, pp. 59–85. Kluwer Academic Publishers, Boston/Dordrecht/London (2002) |
| [35] | IBM: Language extension in C/C++ for cell–version 2.5 (2008) |
| [36] | Lowe D.G.: Distinctive image features from scale-invariant keypoints. Int. J. Comput. Vision 60(2), 91–110 (2004) · doi:10.1023/B:VISI.0000029664.99615.94 |
| [37] | Carmichael, O.: Discriminative techniques for the recognition of complex-shaped objects. Ph.D. dissertation, The Robotics Institute, Carnegie Mellon University (2003) |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
