深度迁移学习在小批量图像分类中的应用

Application of Deep Transfer Learning in Mini-Batch Image Classification

利用深度迁移学习算法,将深度模型迁移至小批量数据中进行使用,解决过拟合和对数据标签依赖性强的问题。首先,将已经训练好的模型应用在相似图像分类任务中,提高模型效率;其次利用微调策略,对深度学习网络全连接层进行调整,丢弃部分神经元以降低过拟合的发生,提高模型准确性;最后使用DogsVSCats数据集进行测试。实验结果表明,深度迁移学习算法在小批量样本数据中具有更高的准确性。

Deep transfer learning algorithm is used to transfer deep models to small batch data for use to solve the problems of overfitting and strong dependence on data labels.Firstly,the trained model is applied to the similar image classification task to improve the efficiency of the model.Then,fine-tuning parameters were used to adjust the fully connected layer of the deep learning network,and neurons were discarded to reduce the occurrence of overfitting and improve the accuracy of the model.Finally,DogsVSCats dataset is used for testing.Experimental results show that the deep transfer learning algorithm has higher accuracy in small batch sample data.