动作识别中基于深度神经网络和GA合并算法的分类决策方法

针对人体动作识别中传统方法在分类决策方面存在问题和缺陷，提出了一种新颖的基于深度神经网络(DNN)和遗传算法(GA)合并算法的非线性分类决策方法。首先，提出的合并算法在整个训练集合上对特征提取器进行组合，进而组合成不同的两个独立网络；再利用DNN对两个独立网络进行初始化，进一步利用GA对两个网络进行合并。然后将网络的偏差和权重表示为每层网络间的一个矩阵；最后，利用DNN对网络的偏差和权重进行训练，并在合并过程中将矩阵中的每一行当作一个染色体。实验采用了标准MNIST数据集对提出算法的性能进行评估。评估结果显示实验过程中的交叉和突变操作增加了神经元节点，提高了识别性能，并且弱化了不相关和相关神经元节点。因此，提出算法的错误率更低，网络性能更优异。

Classification decision method based on depth neural network and GA merging algorithm in motion recognition

Aiming at the problems and shortcomings of traditional methods in human motion recognition in classification decision, a novel nonlinear classification decision method based on deep neural network (DNN) and genetic algorithm (GA) merge algorithm is proposed. First, the proposed merging algorithm combines the feature extractors over the entire training set and combines them into two different independent networks. Then use DNN to initialize two independent networks and further use GA to merge the two networks. Then the deviation and weight of the network are expressed as a matrix between each layer of the network. Finally, use DNN to train the bias and weight of the network, and each row in the matrix is treated as a chromosome during the merge process. The experiment uses the standard MNIST data set to evaluate the performance of the proposed algorithm. The evaluation results showed that the crossover and mutation operations during the experiment increased the neuron nodes, improved the recognition performance, and weakened the irrelevant and related neuronal nodes. Therefore, the proposed algorithm has a lower error rate and better network performance.