Character recognition using a fast neural-net classifier
暂无分享,去创建一个
Abstract A correlation-based neural-net classifier is used to classify printed characters using moment invariants as the feature inputs. This classifier drastically reduces the training time by setting up initial weights and a hidden layer structure more suitable for fast convergence.
[1] Richard P. Lippmann,et al. An introduction to computing with neural nets , 1987 .
[2] Eric A. Wan,et al. Neural network classification: a Bayesian interpretation , 1990, IEEE Trans. Neural Networks.
[3] Ming-Kuei Hu,et al. Visual pattern recognition by moment invariants , 1962, IRE Trans. Inf. Theory.
[4] Yoh-Han Pao,et al. Adaptive pattern recognition and neural networks , 1989 .