Neural architecture search (NAS) has significant progress in improving the accuracy of image classification. Recently, some works attempt to extend NAS to image segmentation which shows preliminary feasibility. However, all of them focus on searching architecture for semantic segmentation in natural scenes. In this paper, we design three types of primitive operation set on search space to automatically find two cell architecture DownSC and UpSC for semantic image segmentation especially medical image segmentation. Inspired by the U-net architecture and its variants successfully applied to various medical image segmentation, we propose NAS-Unet which is stacked by the same number of DownSC and UpSC on a U-like backbone network. The architectures of DownSC and UpSC updated simultaneously by a differential architecture strategy during the search stage. We demonstrate the good segmentation results of the proposed method on Promise12, Chaos, and ultrasound nerve datasets, which collected by magnetic resonance imaging, computed tomography, and ultrasound, respectively. Without any pretraining, our architecture searched on PASCAL VOC2012, attains better performances and much fewer parameters (about 0.8M) than U-net and one of its variants when evaluated on the above three types of medical image datasets.INDEX TERMS Medical image segmentation, convolutional neural architecture search, deep learning.

With the rapid development of Internet technology and social networks, a large number of comment texts are generated on the Web. In the era of big data, mining the emotional tendency of comments through artificial intelligence technology is helpful for the timely understanding of network public opinion. The technology of sentiment analysis is a part of artificial intelligence, and its research is very meaningful for obtaining the sentiment trend of the comments. The essence of sentiment analysis is the text classification task, and different words have different contributions to classification. In the current sentiment analysis studies, distributed word representation is mostly used. However, distributed word representation only considers the semantic information of word, but ignore the sentiment information of the word. In this paper, an improved word representation method is proposed, which integrates the contribution of sentiment information into the traditional TF-IDF algorithm and generates weighted word vectors. The weighted word vectors are input into bidirectional long short term memory (BiLSTM) to capture the context information effectively, and the comment vectors are better represented. The sentiment tendency of the comment is obtained by feedforward neural network classifier. Under the same conditions, the proposed sentiment analysis method is compared with the sentiment analysis methods of RNN, CNN, LSTM, and NB. The experimental results show that the proposed sentiment analysis method has higher precision, recall, and F 1 score. The method is proved to be effective with high accuracy on comments. INDEX TERMS Sentiment analysis, artificial intelligence, social network, weighted word vectors, BiLSTM.

With the rapid development of the Internet, the amount of data has grown exponentially. On the one hand, the accumulation of big data provides the basic support for artificial intelligence. On the other hand, in the face of such huge data information, how to extract the knowledge of interest from it has become a matter of general concern. Topic tracking can help people to explore the process of topic development from the huge and complex network texts information. By effectively organizing large-scale news documents, a method for the evolution of news topics over time is proposed in this paper to realize the tracking and evolution of topics in the news text set. First, the LDA (latent Dirichlet allocation) model is used to extract topics from news texts and the Gibbs Sampling method is used to speculate parameters. The topic mining using the K-means method is compared to highlight the advantages of using LDA for topic discovery. Second, the improved single-pass algorithm is used to track news topics. The JS (Jensen-Shannon) divergence is used to measure the topic similarity, and the time decay function is introduced to improve the similarity between topics with the similar time. Finally, the strength of the news topic and the content change of the topic in different time windows are analyzed. The experiments show that the proposed method can effectively detect and track the topic and clearly reflect the trend of topic evolution.

No abstract

With the rapid development of the Internet, more and more users expressed their views on the Internet. Therefore, the big data of texts are generated on the Internet. In the era of big data, mining the sentiment tendencies contained in massive texts on the Internet through natural language processing technology has become an important way of public opinion supervision. In this paper, the sensitive information topics-based sentiment analysis method for big data is proposed. This method integrates topic semantic information into text representation through a neural network model. The attention mechanism is introduced into the neural network, and context-aware vector is introduced to calculate the weight of each word. In addition, in order to make the model more adaptable, the method of sentiment dictionary tagging is used to obtain the training data. The experimental results show that the proposed model can effectively improve the accuracy of sentiment analysis results.