Abstract
We developed a machine vision system around an analog neural net chip and used it in several applications. Some of them were: locating the address blocks on mail pieces, finding the identification numbers on rail cars, and discriminating between handwritten and machine-printed characters. The chip, operating as a coprocessor of a workstation, provides a speed-up of a factor of 1000, compared with the workstation. The computation speed achieved lies between one and ten billion multiply-accumulates/s. The neural net chip is based on building blocks,neurons, that can be arranged in various network architectures. The dataflow is optimized for implementing large, structured neural nets, and is also suited for any task in which signals are to be convolved with many kernels. Some of the networks are trained on the neural net chip with a weight-perturbation learning algorithm that was adapted to work with the coarse quantization of the weights and the states in the chip.
Similar content being viewed by others
References
Alspector J, Alien RB, Jayakumar A, Zeppenfeld T, Meir R (1991) Relaxation networks for large supervised learning problems. In: Touretzky DS, Lippman R (eds) Advances in neural information processing systems 3, Morgan Kaufmann, San Mateo, 1015–1021
Arima Y, Mashiko K, Okada K, Yamada T, Maeda Y, Notani H Kon-doh H, Kayano S (1991) A 336-neuron 28k-synapse self-learning neural-network chip with branch-neuron-unit architecture. IEE Digest of the International Solid State Circuits Conference, San Francisco, pp 182–183
Dembo A, Kailath T (1990) Model-free distributed learning. IEEE Trans Neural Networks 1:58–70
Dobbins A, Zucker SW, Cynader MS (1987) Endstopped Neurons in the visual cortex as a substrate for calculating curvature. Nature 329: 438–441
Frye RC, Rietman EA, Wong CC (1991) Back-propagation learning in neural network hardware. IEEE Trans Neural Networks 2:110–117
Fukushima T (1990) A survey of image processing LSIs in Japan. Proceedings 10th International Copnference on Pattern Recognition, Atlantic City, IEEE Computer Society Press, Los Alamitos, pp 394–401
Graf HP, Henderson D, (1990) A reconflgurable cmos neural network, Digest Integrated Solid State Circuits Conference, San Francisco, IEEE, pp 144–145
Graf HP, Janow R., Henderson D, Lee R (1991) Reconfigurable neural net chip with 32K connections. In: Touretzky DS, Lippman R (eds) Advances in neural information processing systems 3. Morgan Kaufmann, San Mateo, Calif., pp 1032–1038
Graf HP, Janow R, Nohl CR, Ben J (1991) A neural-net board system for machine vision applications. Proceeding of the International Joint Conference on Neural Networks, Seattle, Wash. 1:481–486
Grimson WH (1990) The combinatorics of object recognition in cluttered environments using constraint search. Artif Intell 44:121
Holler M, Tam S, Castro H, Benson (1989) An electrically trainable artificial neural network (ETANN) with 10 240 floating gate synapses. Proceedings International Joint Conference on Neural Networks, Washington DC, pp 191–196
Hubel DH, Wiesel TN (1962) Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. J Physiol Lond 106:154
IEEE (1992) Special Issue on Neural Network Hardware, IEEE Trans Neural Networks 3(3)
Jones JP, Palmer LA (1987) An evaluation of the two-dimensional gabor filter model of simple receptive fields in cat striate cortex. J Neurophysiology 58: 1233–1258
LeCun Y, Boser B, Denker JS, Henderson D, Howard RE, Hubbard W, Jackel LD (1990) Handwritten digit recognition with a back-propagation network. In: Touretzky DS (ed) Advances in neural information processing systems 2. Morgan Kaufmann, San Mateo, pp 396–404
Mehrotra R, Nichani S (1990) Corner detection. Patt Recogn 23:1223–1233
Pelgrom MJ, Wulms HE, Stokker P van der, Bergamaschi RA (1987) FEBRIS: a chip for pattern recognition. IEEE J Solid State Circuits 22:423–429
Ramacher U, Rückert U, Nossek J (eds) Proceedings of the 2nd International Conference on Microelectronics for Neural Networks, Kyrill & Method, Munich
Rangarajan K, Shah M, Van Brackle D (1989) Optimal corner detection. Computer Vision Graph Image Processing 48:230–245
Ruetz PA (1989) The architecture and design of a 20-MHz real-time chip set. IEEE J Solid State Circuits 24:338–348
Säckinger E, Boser BE, Bromley J, LeCun Y, Jackel LD (1992) Application of the ANNA neural network chip to high-speed character recognition. IEEE Trans Neural Networks 3:498–505
Satyanarayana S, Tsividis Y, Graf HP (1992) A reconfigurable VLSI neural network. IEEE J. Solid State Circuits 27:67–81
Shin Y, Sridhar R, Demjanenko V, Palumbo PW, Shrihari SN (1992) A special-purpose content addressable memory chip for real-time image processing. IEEE J Solid State Circuits 27:737–744
Wang CH, Shrihari SN (1988) A framework for object recognition in a visually complex environment and its application to locating address blocks on mail pieces. Int J Comput Vision 2: 119–145
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Graf, H.P., Nohl, C.R. & Ben, J. Image recognition with an analog neural net chip. Machine Vis. Apps. 8, 131–140 (1995). https://doi.org/10.1007/BF01213478
Issue Date:
DOI: https://doi.org/10.1007/BF01213478