说明：  一个神经网络你模式识别的例子--螃蟹识别，模式识别主要解决分类问题，由于分类问题的输出是离散的，而曲线拟合的输出是连续的，因此分类问题的解决方案会与曲线拟合问题的解决方案存在一些差异。
(A neural network example of your pattern recognition - crab recognition, pattern recognition is mainly to solve the classification problem, because the classification problem output is discrete, and the curve fitting output is continuous, so the classification problem solution will be different from the curve fitting problem solution.)