Neocognitron for handwritten digit recognition
Neocognitron for handwritten digit recognition
复制标题
DOI：
10.1016/s0925-2312(02)00614-8
复制
发表时间：
2003-04-01
期刊：
影响因子：
6
通讯作者：
Fukushima, K
中科院分区：
文献类型：
--
作者：
Fukushima, K
作者以前提出了一个神经网络模型neocognitron鲁棒视觉模式识别。本文提出了一种改进的neocognitron版本，并使用大型手写数字数据库（ETLI）证明了其能力。为了提高neocognitron的识别率，进行了几项修改：如S-细胞与C-细胞连接中的抑制环境、输入层与边缘提取层之间的对比提取层、线条提取细胞的自组织、在最高阶段的监督竞争学习，S-和C-细胞的交错排列，等等。这些修改允许删除附加到以前版本的附属电路，从而提高了识别率，简化了网络结构，识别率随训练次数的变化而变化模式.例如，当我们使用3000个数字（每个数字300个模式）进行学习时，盲测试集（3000个数字）的识别率为98.6%，训练集为100%。(C)2002 Elsevier Science B. V.保留所有权利。
The author previously proposed a neural network model neocognitron for robust visual pattern recognition. This paper proposes an improved version of the neocognitron and demonstrates its ability using a large database of handwritten digits (ETLI).To improve the recognition rate of the neocognitron, several modifications have been applied: such as, the inhibitory surround in the connections from S-cells to C-cells, contrast-extracting layer between input and edge-extracting layers, self-organization of line-extracting cells, supervised competitive learning at the highest stage, staggered arrangement of S- and C-cells, and so on. These modifications allowed the removal of accessory circuits that were appended to the previous versions, resulting in an improvement of recognition rate as well as simplification of the network architecture.The recognition rate varies depending on the number of training patterns. When we used 3000 digits (300 patterns for each digit) for the learning, for example, the recognition rate was 98.6% for a blind test set (3000 digits), and 100% for the training set. (C) 2002 Elsevier Science B.V. All rights reserved.