{"title":"Portrait Sketch Synthesis via Mish-Gated U-net and GANs","authors":"Jian Chen, Gang Liu, Gui Yuan, Xinyun Wu","doi":"10.1145/3578741.3578753","DOIUrl":"https://doi.org/10.1145/3578741.3578753","url":null,"abstract":"GANs structure. However, those methods based on GANs did not get a better effect, their results usually have obvious quality problems such as serious distortion and visual artifacts. In this paper, we propose a novel portrait sketch synthesis approach based on the mish-gated U-net (MGUnet) and GANs to address the above problems. The MGUnet is an improvement on the classical U-net, which utilizes mish activation function to build a new gating module, and embeds the module in U-net to enhance the synthesis performance of the neural network. The proposed method can effectively eliminate artifacts and enhance the effect of synthesized images. The results of our experiments demonstrate that our method can convert satisfactory portrait sketches for portrait sketch synthesis.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121346163","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Qian Cheng, Youwei Zhang, Bing Xue, Fangli Ge, Wenqing Yao
{"title":"Relevance calculation of users' document sending behavior under the network social platform","authors":"Qian Cheng, Youwei Zhang, Bing Xue, Fangli Ge, Wenqing Yao","doi":"10.1145/3578741.3578750","DOIUrl":"https://doi.org/10.1145/3578741.3578750","url":null,"abstract":"The correlation calculation of online users' posting behavior is studied as the core part in the analysis and characterization of Internet user entity identity association. The paper will complete the processing of the information posted by the network users and train the bag-of-words model to obtain the feature vector of the posted information, and then use the distance formula to calculate the correlation of the posted behavior information, which can assist in the analysis and characterization of network user entity identity association and associate the identity relationship between multiple virtual users, so as to build a network user relationship network. The method in this paper is experimentally verified on the published data of WeChat public account, using three training models, Word2Vec, CNN and Seq2Seq. The results prove the effectiveness of the text method, and provide research ideas and method support for the analysis and characterization of network user entity identity associations.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"2015 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127260052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Method of Machine Learning for Social Bot Detection Combined with Sentiment Analysis","authors":"Gu Long, Deyu Lin, Jie Lei, Zhiyong Guo, Yangyang Hu, Linglin Xia","doi":"10.1145/3578741.3578790","DOIUrl":"https://doi.org/10.1145/3578741.3578790","url":null,"abstract":"Social Bot exists widely in major social networks. Some maliciously use a social bot to guide public opinion, steal user privacy, and create rumors, which seriously affects the security of social networks. Past approaches mainly extracted large amounts of contents but ignored bots’ text sentiment features, and it is hard to detect social bot just based on contents. This paper proposes a malicious social bot detection method that combines sentiment features in response to this problem. It trains a Bidirectional Long Short-Term Memory model(Bi-LSTM) with an Attention Mechanism to perform sentiment calculation on the online text information of social accounts and analyze the sentiment fluctuations of accounts to get the new sentiment features; Then, it inputs the new features combined with metadata features into different machine learning models for analysis and comparison. Through this method, different machine learning detection models have improved the detection accuracy after combining sentiment features.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127534447","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hao-Yu Duan, Haoyu Jin, Kang Chen, Shaochong Du, T. Fang, H. Huo
{"title":"An effective Time-Aware Encoder for Temporal Knowledge Graph Reasoning","authors":"Hao-Yu Duan, Haoyu Jin, Kang Chen, Shaochong Du, T. Fang, H. Huo","doi":"10.1145/3578741.3578758","DOIUrl":"https://doi.org/10.1145/3578741.3578758","url":null,"abstract":"There are many studies on static knowledge graph (KG) reasoning that predicts missing facts for its completeness. As the facts in real world usually are time-dependent, temporal knowledge graph (TKG) has received great attention lately. Entities and their relations in TKGs may change over time, and how to predict future facts from past facts has become a fundamental subject of TKGs. In this paper, a novel effective time-aware encoder (TAE) is proposed for TKG reasoning. It encodes the influence of time on entities and relations into accurate time-specific embedding representations. Then the embedding representations of entities and relations under different timestamps are employed for the prediction of future facts. The evaluations on four public TKG datasets have demonstrated that TAE outperforms all baseline models on TKG reasoning tasks.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114497518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Faster Method For Generating Chinese Text Summaries-Combining Extractive Summarization And Abstractive Summarization","authors":"Wenchuan Yang, Tianyu Gu, Runqi Sui","doi":"10.1145/3578741.3578752","DOIUrl":"https://doi.org/10.1145/3578741.3578752","url":null,"abstract":"Extractive summarization and generative summarization are the two main ways to generate summarization.However,previous work treats both of them as two independent subtasks.In this paper,we obtain new summarization by combining extractive summarization and generative summarization.This method extracts the key information of the article firstly,and then generates the summarization of the extracted information.The experimental result shows that this method can significantly improve the quality of the generative text compared with extractive summarization,and can significantly improve the generative speed compared with generative summarization.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129223858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Chinese medical named entity recognition based on zero-shot learning","authors":"Menglin Zhou, Kecun Gong","doi":"10.1145/3578741.3578780","DOIUrl":"https://doi.org/10.1145/3578741.3578780","url":null,"abstract":"To solve the problem that existing named entity recognition models lacked of ability to deal with unseen classes, the zero-shot learning was proposed to be used in the task of Chinese medical named entity recognition. Zero-shot learning utilizes the description information of the entity's class to establish the connection between the entity and the class, and transfers information from the observed classes to the unseen target classes. The model proposed by this paper was mainly based on BERT, which was used to model the relationship between the entity and the description. Moreover, the static word embedding, which is as a supplementary information, is concatenated with the features obtained from the BERT to solve the problem that BERT is not suitable for a specific field. At the same time, Correlation Searchers are added between the transformers of BERT to search for the word information most relevant to the character, so as to solve the problem that the model cannot obtain complete word information with characters as the input unit. Experiments show that the model's recognition performance has been significantly improved after adding the static word embedding and word information.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128045915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Keyword Extractor for Contrastive Learning of Unsupervised Sentence Embedding","authors":"Hua Cai, Weihong Chen, Kehuan Shi, Shuaishuai Li, Qing Xu","doi":"10.1145/3578741.3578759","DOIUrl":"https://doi.org/10.1145/3578741.3578759","url":null,"abstract":"Contrastive learning has been widely applied to learning unsupervised sentence embedding. One major method is unsupervised SimCSE, which only uses random dropouts as noise to build positive pairs for contrastive learning. But token-level matching does not always properly represent the similarity between texts. Sentence-level semantic representation based on keyword-level could enhance the similarity matching performance between texts. To emphasize the contribution of keywords in sentence representation, we propose KESimCSE, which not only includes the constrastive learning of sentence embedding but also adds KL divergence of two elements to the loss function. One element is the dot product of each token embedding and the [CLS] embedding of the BERT outputs, as [CLS] generally captures the semantics of the whole sentence. The other element is the weight list of all tokens obtained through keyword extraction method. Our experiments show that the averaged Spearman’s correlation of KESimCSE on semantic textual similarity tasks raises to 77.21%, which outperforms unsupervised SimCSE by nearly 1.30%.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133773420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Pixel Classification of Hyperspectral Images based on Spectral and Gabor Texture Features","authors":"Bingchun Li, Dong-ping Tian, J. Liu, Yali Wang","doi":"10.1145/3578741.3578800","DOIUrl":"https://doi.org/10.1145/3578741.3578800","url":null,"abstract":"Hyperspectral images consist of rich spectral information and spatial structure information that can be fully employed to implement the ground objects classification effectively. Due to the strong representation ability of Gabor filter to extract relevant features at different scales and directions, this paper proposes a Gabor filter group with 5 wavelengths and 4 directions that is exploited to the two-dimensional image of each band. To be specific, the 3×3 local neighborhood mean square error of the filtered image pixel is used as the texture feature of the pixel in a certain filtered image, and the texture features of all the filtered images in each band are combined together as the overall texture features of this pixel. XGBoost is leveraged as the classifier to carry out the classification experiments based on spectral feature, Gabor texture feature, and the combined features of the spectral and Gabor texture features, respectively. The experimental results show that our proposal can obtain the accuracy of , and based on only used spectral features, only used texture features, and the combined features, respectively. It is obvious that the classification accuracy can be significantly enhanced based on the combined features of the spectral and Gabor texture features.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116958195","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Rong Wu, Huifang Huang, Beibei Liu, Jianhua Gao, Lan Wei, Xiaolu Fei
{"title":"Chinese vascular ultrasound report Generation based on LSTM-RNN-Attention model","authors":"Rong Wu, Huifang Huang, Beibei Liu, Jianhua Gao, Lan Wei, Xiaolu Fei","doi":"10.1145/3578741.3578834","DOIUrl":"https://doi.org/10.1145/3578741.3578834","url":null,"abstract":"It is a tedious task to summarize the diagnostic impressions from the image description of vascular ultrasound, and it is more possible to make mistakes because it requires rich experience and clear logical thinking ability of ultrasound radiologists. In this paper, we explore the use of a LSTM-RNN-Attention (LRA) model to automatically generate diagnostic impressions based on the description of vascular ultrasound images, and evaluate the performance of the model by ROUGE scores. Also, the paper compares this model with pointer generation network (PGN) and basic encoder-decoder (ED) model. The LRA Model achieved a ROUGE-1 score of 26.353, a ROUGE-2 score of 11.587, a ROUGE-L score of 26.168, each score was almost 8 times higher than it of (PGN). The results showed that LRA model presented better performance than these most commonly used models in the field of chinses radiological text summarization and is useful in clinical practice.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"5 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117327820","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The Application of Natural Language Processing and tspDB to Predict Future Bitcoin Price","authors":"Kaige Bao","doi":"10.1145/3578741.3578798","DOIUrl":"https://doi.org/10.1145/3578741.3578798","url":null,"abstract":"Formerly, predictions of time series were usually based on numbers. In some cases, however, one has to make predictions based on linguistic information. In this paper, we experiment with time series forecasting based on textual information. We combine sentiment analysis with time series forecasting to predict fluctuations in variables that change over time. Due to the time-sensitivity of the target variable, models from the Time Series Prediction Database (tspDB) are used to infer the value of Bitcoin. We implemented an innovative architecture that accepts text as input and produces numerical predictions for certain values. The market value of bitcoin was used to verify the applicability of the architecture. The program inputs Twitter posts and outputs the market value of bitcoin for the next few days. The program first calculates a sentiment score (which reflects social media confidence in the price of bitcoin) and then makes a multivariate time series prediction of the price of bitcoin. The results can provide some insight into how the price of bitcoin fluctuates. The purpose of using Bitcoin as our primary benchmark is that we want to demonstrate a sound application of the combination of NLP and tspDB, and while Bitcoin price is not an objective variable, the method we apply can also be used in other situations that contain linguistic information as input.","PeriodicalId":306892,"journal":{"name":"Proceedings of the 2022 5th International Conference on Machine Learning and Natural Language Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129431265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}