Cellular Neural Network training by ant colony optimization algorithm

Cellular Neural Networks (CNN) having parallel processing capabilities present important advantages in image processing applications. The coefficients of the template matrices and the threshold values of CNN should be optimized to obtain the desired output image. The learning algorithms designed for classical feed forward neural networks are not suitable for CNN due to its dynamic architecture. Researchers are still working on development of generalized learning algorithms for CNN. In this study, the CNN training is realized by ant colony optimization (ACO) technique. The results obtained by trained CNN show that ant colony based learning algorithm is very successful for image feature extraction problems such as edge, corner, vertical and horizontal edge detections.

[1]  G. Theraulaz,et al.  Inspiration for optimization from social insect behaviour , 2000, Nature.

[2]  Marco Dorigo,et al.  Distributed Optimization by Ant Colonies , 1992 .

[3]  Eugenio Di Sciascio,et al.  A new learning algorithm for pattern classification using cellular neural networks , 2001, ISCAS 2001. The 2001 IEEE International Symposium on Circuits and Systems (Cat. No.01CH37196).

[4]  Yoshifumi Nishio,et al.  Cellular neural network with dynamic template and its output characteristics , 2009, 2009 International Joint Conference on Neural Networks.

[5]  Mta Sztaki.  Texture Classification and Segmentation by Cellular Neural Networks Using Genetic Learning , 1998 .

[6]  Xin Yao,et al.  Assignment of cells to switches in a cellular mobile network using a hybrid Hopfield network-genetic algorithm approach , 2008, Appl. Soft Comput..

[7]  C. Guzelis,et al.  Recurrent perceptron learning algorithm for CNNs with application to edge detection , 1995 .

[8]  Akio Ushida,et al.  Adaptive Simulated Annealing in CNN Template Learning , 1999 .

[9]  Zhigang Zeng,et al.  Associative memories based on continuous-time cellular neural networks designed using space-invariant cloning templates , 2009, Neural Networks.

[10]  Cuneyt Guzelis,et al.  Image restoration using cellular neural network , 1997 .

[11]  Lin-Bao Yang,et al.  Cellular neural networks: theory , 1988 .

[12]  Z. Bingul,et al.  A new PID tuning technique using ant algorithm , 2004, Proceedings of the 2004 American Control Conference.

[13]  Girolamo Fornarelli,et al.  Adaptive particle swarm optimization for CNN associative memories design , 2009, Neurocomputing.

[14]  Abdullah Bal,et al.  Wavelet-cellular neural network architecture and learning algorithm , 2004, SPIE Defense + Commercial Sensing.

[15]  Giuseppe Grassi,et al.  Learning algorithm for pattern classification using cellular neural networks , 2000 .

[16]  Luca Maria Gambardella,et al.  Ant Algorithms for Discrete Optimization , 1999, Artificial Life.

[17]  Abdullah Bal.  Widrow-cellular neural network and optoelectronic implementation , 2004 .

[18]  Enis Günay,et al.  Efficient edge detection in digital images using a cellular neural network optimized by differential evolution algorithm , 2009, Expert Syst. Appl..

[19]  T. Ogura,et al.  CAM2-universal machine: A DTCNN implementation for real-time image processing , 2008, 2008 11th International Workshop on Cellular Neural Networks and Their Applications.