【摘要】 文章提出了模式识别的最大熵方法,其基本思想是求出最大熵概率分布,再求出条件概率分布,进而作出二值分类。它的特点是能最大限度地利用已有信息做出最合理的推测。求解最大熵分布时,需要解复杂的约束优化问题,为此使用了神经网络,从而使该方法结合了神经网络的很多优点。该方法的突出优点是在小样本情况下仍能保持很好的识别率。更多还原

【Abstract】 In this paper,maximum entropy method is applied to pattern recognition,the key point is to find the proba-bility distribution that has maximum entropy measure,which ensures that the distribution is most reasonable.Solving the maximum entropy distribution involves a constrained optimization,and by constructing a neural network,the algorithm is more robust.This method is especially useful when the sample set is small.更多还原