Abstract:

In pattern recognition, samples are directly classified in the feature space or mapped to higher dimensional spaces by kernel functions. In order to improve the classification effect, better separation surface and more efficient feature space are commonly used. This paper proposes a Dynamic Hypersphere Algorithm（DHA）  for classification that can find a more efficient feature space. DHA uses dynamic feature transformation and optimizes the same type of data into the same hypersphere. This paper proves that DHA can obtain better classification results by finding effective feature space by experimenting on standard data sets. In addition, in order to further prove the validity of the feature space, DHA is applied to the MNIST handwriting by reducing the training samples and reducing the original sample from 784 dimensions to 10 dimensions, a recognition rate of 90.18% is obtained, and good results are also obtained in the unbalanced handwriting.

Key words: neural network, dynamic feature space, dynamic hypersphere, standard data sets, MNIST handwriting

摘要：

在模式识别中，样本在特征空间被直接用作分类或被核函数映射至更高维空间分类。为了提高分类效果，一般使用更好的分隔面和更有效的特征空间。提出了一种使用神经网络寻找到更有效特征空间的动态超球体算法（Dynamic Hypersphere Algorithm，DHA）。DHA采用了动态的特征变换，通过满足优化超球体的条件获得更有效的特征空间，最终通过欧氏距离得到分类结果。在标准数据集上实验证明了DHA能够通过动态的特征变换寻找到有效特征空间，从而获得更好的分类效果。为了进一步证明特征空间的有效性，将DHA应用到MNIST手写体，通过减少训练样本并且将原样本由784维降至10维获得了90.18%的识别率，在不平衡手写体中也获得了较好的效果。

关键词: 神经网络, 动态特征空间, 动态超球体, 标准数据集, MNIST手写体