Browsing by Author "Lai, JL"

Now showing 1 - 3 of 3

Architectural design and analysis of learnable self-feedback ratio-memory cellular nonlinear network (SRMCNN) for nanoelectronic systems
(2004-11-01)
Lai, JL
;
Wu, PCY
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
In this paper, a learnable cellular nonlinear network (CNN) with space-variant templates, ratio memory (RM), and modified Hebbian learning algorithm is proposed and analyzed. By integrating both the modified Hebbian learning algorithm with the self-feedback function and a ratio memory into CNN architecture, the resultant ratio-memory (RMCNN) is called the self-feedback RMCNN (SRMCNN) which can serve as the associative memory. It can generate the absolute weights and then transform them into the ratioed A-template weights as the ratio memories for recognizing noisy input patterns. Simulation results have shown that with the stronger feature enhancement effect, the SRMCNN under constant leakage current can store and recognize more patterns than the RMCNN. For 18 x 18 SRMCNN, 93 noisy patterns with a uniform distribution noise level of 0.8 and a variance of normal distribution noise of 0.3 can be learned, stored, and recognized with 100% success rate. The SRMCNN has greater learning and recognition capability when the learned patterns are simpler and the noise is lower. For the learning and recognition of complicated patterns, the allowable pattern number is decreased for a 100% success rate. Simulation results have successfully verified the correct functions and better performance of SRMCNN in the pattern recognition. With high integration capability and excellent pattern association performance, the proposed SRMCNN can be applied to nanoelectronic associative-memory systems for image processing applications.
Improvement of pattern learning and recognition capability in ratio-memory cellular neural networks with non-discrete-type Hebbian learning algorithm
(2002)
Wu, CY
;
Lai, JL
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
A ratio-memory cellular neural networks (RMCNN) with non-discrete-type Hebbian teaming algorithm to learn and recognize the image patterns is proposed and analyzed. In the proposed RMCNN, the space-variant A templates with self-feedback coefficients are determined from the trained patterns using the non-discrete-type Hebbian teaming algorithm during the teaming period. The determined A templates stored in the ratio memory are used in the RMCNN to recognize the learned patterns with different Gaussian noise levels and output the correct patterns. The operation of the proposed RMCNN has been simulated with Matlab software. It is shown that the 9x9 RMCNN can successfully learn recognize 23 noisy patterns with Gaussian noise variance of 0.3. As compared to other learnable CNNs as associate memories, the proposed RMCNN with non-discrete-type Hebbian teaming algorithm and 5 coefficients in A template can learn and recognize much more patterns. With improved pattern teaming and recognition capability, the proposed RMCNN still can be implemented in VLSI for various applications.
Learnable self-feedback ratio-memory cellular nonlinear network (SRMCNN) for associative memory applications
(2004)
Lai, JL
;
Wu, Y
;
電子工程學系及電子研究所
;
Department of Electronics Engineering and Institute of Electronics
A self-feedback ratio-memory cellular nonlinear network (SRMCNN) with the B template and the modified Hebbian learning algorithm to learn and recognize the image patterns is proposed and analyzed. In the proposed SRMCNN, the coefficients of space-variant B templates are determined from the exemplar patterns during the learning period. The weights are the ratio of the absolute summation of its neighborhood weights in the B templates was stored in the associative memory. This SRMCNN can recognize the learned patterns with distinct white-black noise and output the correct patterns. The Matlab and HSPICE software has been simulated the operation of the proposed SRMCNN. It is shown that the 18x18 SRMCNN can successfully learned and recognized 8 incompletely noisy patterns. As compared to other learnable CNN as associate memories, the proposed SRMCNN could improve pattern learning and recognition capability. The architecture can be implemented in nano-CMOS technology for giga-scale learning system in the real-time applications.