H. Huynh, Tin Trung Dang, Linh My Thi Ong, H. H. Luong, Nghia Duong-Trung, Tai Tan Phan, B. Pottier
A nation-wide concern for the sustainability of mangrove forests in Mekong Delta (Vietnam) is increasingly recognized. Unfortunately, overexploitation of natural resources, urbanization, deforestation, agriculture, aquaculture and many other threats have caused a severe reduction of mangrove cover. Mangrove forests significantly contribute to the provision of local ecosystems and the Delta's sustainability, although they cover a small proportion of the Delta's surface. Therefore, the rehabilitation of mangrove forests demands strong coordinated efforts in terms of policy and research. This study evaluates the potential for simulating mangroves rehabilitation via cellular automata, e.g. a discrete dynamical system, by characterizing several environmental factors. Two of the largest environmental effects causing the distribution of species in mangrove forests are leaf area index (LAI) and flood-tide. To the best of own research, the applied methodologies are one of the first endeavors that have been investigated in the literature. The research has been conducted at Ong Trang islet, Ca Mau province, Mekong Delta (Vietnam).
{"title":"Simulating Mangroves Rehabilitation with Cellular Automata","authors":"H. Huynh, Tin Trung Dang, Linh My Thi Ong, H. H. Luong, Nghia Duong-Trung, Tai Tan Phan, B. Pottier","doi":"10.1145/3380688.3380696","DOIUrl":"https://doi.org/10.1145/3380688.3380696","url":null,"abstract":"A nation-wide concern for the sustainability of mangrove forests in Mekong Delta (Vietnam) is increasingly recognized. Unfortunately, overexploitation of natural resources, urbanization, deforestation, agriculture, aquaculture and many other threats have caused a severe reduction of mangrove cover. Mangrove forests significantly contribute to the provision of local ecosystems and the Delta's sustainability, although they cover a small proportion of the Delta's surface. Therefore, the rehabilitation of mangrove forests demands strong coordinated efforts in terms of policy and research. This study evaluates the potential for simulating mangroves rehabilitation via cellular automata, e.g. a discrete dynamical system, by characterizing several environmental factors. Two of the largest environmental effects causing the distribution of species in mangrove forests are leaf area index (LAI) and flood-tide. To the best of own research, the applied methodologies are one of the first endeavors that have been investigated in the literature. The research has been conducted at Ong Trang islet, Ca Mau province, Mekong Delta (Vietnam).","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123862218","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}
Quang H. Nguyen, R. Muthuraman, Laxman Singh, Gopa Sen, An Tran, Binh P. Nguyen, M. Chua
Diabetic Retinopathy (DR) is an eye disease associated with chronic diabetes. DR is the leading cause of blindness among working aged adults around the world and estimated it may affect more than 93 million people. Progression to vision impairment can be slowed or controlled if DR is detected in time, however this can be difficult as the disease often shows few symptoms until it is too late to provide effective treatment. Currently, detecting DR is a time-consuming and manual process, which requires an ophthalmologist or trained clinician to examine and evaluate digital color fundus photographs of the retina, to identify DR by the presence of lesions associated with the vascular abnormalities caused by the disease. The automated method of DR screening will speed up the detection and decision-making process, which will help to control or manage DR progression. This paper presents an automated classification system, in which it analyzes fundus images with varying illumination and fields of view and generates a severity grade for diabetic retinopathy (DR) using machine learning models such as CNN, VGG-16 and VGG-19. This system achieves 80% sensitivity, 82% accuracy, 82% specificity, and 0.904 AUC for classifying images into 5 categories ranging from 0 to 4, where 0 is no DR and 4 is proliferative DR.
{"title":"Diabetic Retinopathy Detection using Deep Learning","authors":"Quang H. Nguyen, R. Muthuraman, Laxman Singh, Gopa Sen, An Tran, Binh P. Nguyen, M. Chua","doi":"10.1145/3380688.3380709","DOIUrl":"https://doi.org/10.1145/3380688.3380709","url":null,"abstract":"Diabetic Retinopathy (DR) is an eye disease associated with chronic diabetes. DR is the leading cause of blindness among working aged adults around the world and estimated it may affect more than 93 million people. Progression to vision impairment can be slowed or controlled if DR is detected in time, however this can be difficult as the disease often shows few symptoms until it is too late to provide effective treatment. Currently, detecting DR is a time-consuming and manual process, which requires an ophthalmologist or trained clinician to examine and evaluate digital color fundus photographs of the retina, to identify DR by the presence of lesions associated with the vascular abnormalities caused by the disease. The automated method of DR screening will speed up the detection and decision-making process, which will help to control or manage DR progression. This paper presents an automated classification system, in which it analyzes fundus images with varying illumination and fields of view and generates a severity grade for diabetic retinopathy (DR) using machine learning models such as CNN, VGG-16 and VGG-19. This system achieves 80% sensitivity, 82% accuracy, 82% specificity, and 0.904 AUC for classifying images into 5 categories ranging from 0 to 4, where 0 is no DR and 4 is proliferative DR.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"96 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115709070","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}
In this paper, we propose distance-based mean filter (DBMF) to remove the salt and pepper noise. Although DBMF also uses the adaptive conditions like AMF, it uses distance-based mean instead of median. The distance-based mean focuses on similarity of pixels based on distance. It also skips noisy pixels from evaluating new gray value. Hence, DBMF works more effectively than AMF. In the experiments, we test on 20 images of the MATLAB library with various noise levels. We also compare denoising results of DBMF with other similar denoising methods based on the peak signal-to-noise ratio and the structure similarity metrics. The results showed that DBMF can effectively remove noise with various noise levels and outperforms other methods.
{"title":"Distance-Based Mean Filter for Image Denoising","authors":"N. M. Hong, Nguyen Thanh","doi":"10.1145/3380688.3380704","DOIUrl":"https://doi.org/10.1145/3380688.3380704","url":null,"abstract":"In this paper, we propose distance-based mean filter (DBMF) to remove the salt and pepper noise. Although DBMF also uses the adaptive conditions like AMF, it uses distance-based mean instead of median. The distance-based mean focuses on similarity of pixels based on distance. It also skips noisy pixels from evaluating new gray value. Hence, DBMF works more effectively than AMF. In the experiments, we test on 20 images of the MATLAB library with various noise levels. We also compare denoising results of DBMF with other similar denoising methods based on the peak signal-to-noise ratio and the structure similarity metrics. The results showed that DBMF can effectively remove noise with various noise levels and outperforms other methods.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"3 4","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120999833","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}
The stock market price prediction is a challenging real world problem as the prediction model is trained on data with uncertainties and fluctuations. This paper is an attempt to find a membership function with least error of prediction for a fuzzified RIPPER hybrid model, for stock market prediction. The stock market prices were predicted using a hybrid model of FRBS and RIPPER. Three different membership functions of the FRBS, namely triangle, trapezoidal and Gaussian, are considered in this study. The parameters of this function are designed to predict the stock market prices and then MAPE is calculated to determine the membership function that gives the least error. This hybrid model was used to predict the stock prices of four datasets and the MAPE error was calculated for all the membership functions.
{"title":"A Study on the Effect of Fuzzy Membership Function on Fuzzified RIPPER for Stock Market Prediction","authors":"Annie Biby Rapheal, Sujoy Bhattacharya","doi":"10.1145/3380688.3380716","DOIUrl":"https://doi.org/10.1145/3380688.3380716","url":null,"abstract":"The stock market price prediction is a challenging real world problem as the prediction model is trained on data with uncertainties and fluctuations. This paper is an attempt to find a membership function with least error of prediction for a fuzzified RIPPER hybrid model, for stock market prediction. The stock market prices were predicted using a hybrid model of FRBS and RIPPER. Three different membership functions of the FRBS, namely triangle, trapezoidal and Gaussian, are considered in this study. The parameters of this function are designed to predict the stock market prices and then MAPE is calculated to determine the membership function that gives the least error. This hybrid model was used to predict the stock prices of four datasets and the MAPE error was calculated for all the membership functions.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115828487","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}
Emotion recognition plays a particularly important role in the field of artificial intelligence. However, the emotional recognition of electroencephalogram (EEG) in the past was only a unimodal or a bimodal based on EEG. This paper aims to use deep learning to perform emotional recognition based on the multimodal with valence-arousal dimension of EEG, peripheral physiological signals, and facial expressions. The experiment uses the complete data of 18 experimenters in the Database for Emotion Analysis Using Physiological Signals (DEAP) to classify the EEG, peripheral physiological signals and facial expression video in unimodal and multimodal fusion. The experiment demonstrates that Multimodal fusion's accuracy is excelled that in unimodal and bimodal fusion. The multimodal compensates for the defects of unimodal and bimodal information sources.
{"title":"Valence-Arousal Model based Emotion Recognition using EEG, peripheral physiological signals and Facial Expression","authors":"Qi Zhu, G. Lu, Jingjie Yan","doi":"10.1145/3380688.3380694","DOIUrl":"https://doi.org/10.1145/3380688.3380694","url":null,"abstract":"Emotion recognition plays a particularly important role in the field of artificial intelligence. However, the emotional recognition of electroencephalogram (EEG) in the past was only a unimodal or a bimodal based on EEG. This paper aims to use deep learning to perform emotional recognition based on the multimodal with valence-arousal dimension of EEG, peripheral physiological signals, and facial expressions. The experiment uses the complete data of 18 experimenters in the Database for Emotion Analysis Using Physiological Signals (DEAP) to classify the EEG, peripheral physiological signals and facial expression video in unimodal and multimodal fusion. The experiment demonstrates that Multimodal fusion's accuracy is excelled that in unimodal and bimodal fusion. The multimodal compensates for the defects of unimodal and bimodal information sources.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125772866","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}
Current developments in technologies occupy a central role in weather forecasting and the Internet-of-Things for both organizations and the IT sector. Big-data analytics and the classification of data (derived from many sources including importantly the Internet-of-Things) provides significant information on which organizations can optimize their current and future business planning. This paper considers convolutional neural networks and data classification as it relates to big-data and presents a novel approach to weather forecasting. The proposed approach targets the enhancement of convolutional neural networks and data classification to enable improved classification performance for big-data classifiers. Our contribution combines the positive benefits of convolutional neural networks with expert knowledge represented by fuzzy rules for prepared data sets in time series, the aim being to achieve improvements in the predictive quality of weather forecasting. Experimental testing demonstrates that the proposed enhanced convolutional network approach achieves a high level of accuracy in weather forecasting when compared to alternative methods evaluated.
{"title":"Enhancement of Convolutional Neural Networks Classifier Performance in the Classification of IoT Big Data","authors":"Eloanyi Samson Amaechi, H. Pham","doi":"10.1145/3380688.3380702","DOIUrl":"https://doi.org/10.1145/3380688.3380702","url":null,"abstract":"Current developments in technologies occupy a central role in weather forecasting and the Internet-of-Things for both organizations and the IT sector. Big-data analytics and the classification of data (derived from many sources including importantly the Internet-of-Things) provides significant information on which organizations can optimize their current and future business planning. This paper considers convolutional neural networks and data classification as it relates to big-data and presents a novel approach to weather forecasting. The proposed approach targets the enhancement of convolutional neural networks and data classification to enable improved classification performance for big-data classifiers. Our contribution combines the positive benefits of convolutional neural networks with expert knowledge represented by fuzzy rules for prepared data sets in time series, the aim being to achieve improvements in the predictive quality of weather forecasting. Experimental testing demonstrates that the proposed enhanced convolutional network approach achieves a high level of accuracy in weather forecasting when compared to alternative methods evaluated.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127376189","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}
With a wealth of information on hand from the Internet, customers now can easily identify and switch to alternatives. In addition to this, a consensus has been reached that the cost of securing new customers is substantially higher than the cost of retaining the current customers. Therefore, customer retention has become an essential part of operating strategy for any organisation. Churn prediction is a practice of data analysis on the historical data, which is aiming to predict if a customer will be leaving the business or not in advance. A wide range of algorithms have been proposed for churn prediction in the past, however there is no agreement on choosing the best one. Therefore, this study presents a comparative study of the most widely used classification methods on the problem of customer churning in the telecommunication sector. The main goal of this study is to analyse and benchmark the performance of some widely used classification algorithms on a public dataset.
{"title":"Churn Prediction using Ensemble Learning","authors":"Xing Wang, Khang Nguyen, Binh P. Nguyen","doi":"10.1145/3380688.3380710","DOIUrl":"https://doi.org/10.1145/3380688.3380710","url":null,"abstract":"With a wealth of information on hand from the Internet, customers now can easily identify and switch to alternatives. In addition to this, a consensus has been reached that the cost of securing new customers is substantially higher than the cost of retaining the current customers. Therefore, customer retention has become an essential part of operating strategy for any organisation. Churn prediction is a practice of data analysis on the historical data, which is aiming to predict if a customer will be leaving the business or not in advance. A wide range of algorithms have been proposed for churn prediction in the past, however there is no agreement on choosing the best one. Therefore, this study presents a comparative study of the most widely used classification methods on the problem of customer churning in the telecommunication sector. The main goal of this study is to analyse and benchmark the performance of some widely used classification algorithms on a public dataset.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"109 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131585079","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}
Aiming at the problem of unclear description of matching relationship between complex surfaces and insufficient consideration of feature attributes reflecting process constraints, an accurate reasoning algorithm based on topological ranking of feature matching is proposed. Based on the analysis of matching process correlation and feature attributes of typical complex surfaces, a matching topological sequencing model is constructed. The matching objects are encapsulated and topologically ranked by directed acyclic graph, and the precise reasoning among matching feature attributes is realized by using the fuzzy ideal solution. Finally, the validity of the model and the algorithm is verified by an application case.
{"title":"Reasoning Algorithms for Complex Matching Features","authors":"Wei Jiang, Yanchao Yin","doi":"10.1145/3380688.3380689","DOIUrl":"https://doi.org/10.1145/3380688.3380689","url":null,"abstract":"Aiming at the problem of unclear description of matching relationship between complex surfaces and insufficient consideration of feature attributes reflecting process constraints, an accurate reasoning algorithm based on topological ranking of feature matching is proposed. Based on the analysis of matching process correlation and feature attributes of typical complex surfaces, a matching topological sequencing model is constructed. The matching objects are encapsulated and topologically ranked by directed acyclic graph, and the precise reasoning among matching feature attributes is realized by using the fuzzy ideal solution. Finally, the validity of the model and the algorithm is verified by an application case.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129229014","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}
Regression models such as polynomial regression when deployed for training on training instances may sometimes not optimize well and leads to poor generalization on new training instances due to high bias or underfitting due to small value of polynomial degree and may lead to high variance or overfitting due to high degree of polynomial fitting degree. The hypothesis curve is not able to fit all the training instances with a smaller degree due to the changing curvature of curve again and again and also due to the increasing and decreasing nature of curve arising from the local extremas from the plot of points of the dataset curve. The local extremas in between the curve makes the hypothesis curve difficult to fit through all the training instances due to the small polynomial degree. Better optimization and generalization can be achieved by breaking the hypothesis curve into extremas i.e. local maximas and local minimas and deploying separate regression models for each maxima-minima or minima-maxima interval. The number of training instances used to fit the model can be reduced due to very less change in curvature of the curve between an interval due to absence of any local extrema. The time taken by the algorithm reduces due to reduction in the training instances to train which makes the model very less computationally expensive. The algorithm when tested on the UCI machine learning repository datasets gave an accuracy of 53.47% using polynomial regression and 92.06% using our algorithm on Combined Cycle Power Plant Data Set [1] and accuracy of 85.41% using polynomial regression and 96.33% by our algorithm on Real estate valuation Data Set [2]. The approach can be very beneficial for any betterment of mathematical field of study related to bias-variance, cost minimization and better fitting of curves in statistics.
{"title":"Regression Model for Better Generalization and Regression Analysis","authors":"Mohiuddeen Khan, Kanishk Srivastava","doi":"10.1145/3380688.3380691","DOIUrl":"https://doi.org/10.1145/3380688.3380691","url":null,"abstract":"Regression models such as polynomial regression when deployed for training on training instances may sometimes not optimize well and leads to poor generalization on new training instances due to high bias or underfitting due to small value of polynomial degree and may lead to high variance or overfitting due to high degree of polynomial fitting degree. The hypothesis curve is not able to fit all the training instances with a smaller degree due to the changing curvature of curve again and again and also due to the increasing and decreasing nature of curve arising from the local extremas from the plot of points of the dataset curve. The local extremas in between the curve makes the hypothesis curve difficult to fit through all the training instances due to the small polynomial degree. Better optimization and generalization can be achieved by breaking the hypothesis curve into extremas i.e. local maximas and local minimas and deploying separate regression models for each maxima-minima or minima-maxima interval. The number of training instances used to fit the model can be reduced due to very less change in curvature of the curve between an interval due to absence of any local extrema. The time taken by the algorithm reduces due to reduction in the training instances to train which makes the model very less computationally expensive. The algorithm when tested on the UCI machine learning repository datasets gave an accuracy of 53.47% using polynomial regression and 92.06% using our algorithm on Combined Cycle Power Plant Data Set [1] and accuracy of 85.41% using polynomial regression and 96.33% by our algorithm on Real estate valuation Data Set [2]. The approach can be very beneficial for any betterment of mathematical field of study related to bias-variance, cost minimization and better fitting of curves in statistics.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125255524","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}
Multimodal sentiment analysis is still a promising area of research, which has many issues needed to be addressed. Among them, extracting reasonable unimodal features and designing a robust multimodal sentiment analysis model is the most basic problem. This paper presents some novel ways of extracting sentiment features from visual, audio and text, furthermore use these features to verify the multimodal sentiment analysis model based on multi-head attention mechanism. The proposed model is evaluated on Multimodal Opinion Utterances Dataset (MOUD) corpus and CMU Multi-modal Opinion-level Sentiment Intensity (CMU-MOSI) corpus for multimodal sentiment analysis. Experimental results prove the effectiveness of the proposed approach. The accuracy of the MOUD and MOSI datasets is 90.43% and 82.71%, respectively. Compared to the state-of-the-art models, the improvement of the performance are approximately 2 and 0.4 points.
{"title":"Multimodal sentiment analysis based on multi-head attention mechanism","authors":"Chen Xi, G. Lu, Jingjie Yan","doi":"10.1145/3380688.3380693","DOIUrl":"https://doi.org/10.1145/3380688.3380693","url":null,"abstract":"Multimodal sentiment analysis is still a promising area of research, which has many issues needed to be addressed. Among them, extracting reasonable unimodal features and designing a robust multimodal sentiment analysis model is the most basic problem. This paper presents some novel ways of extracting sentiment features from visual, audio and text, furthermore use these features to verify the multimodal sentiment analysis model based on multi-head attention mechanism. The proposed model is evaluated on Multimodal Opinion Utterances Dataset (MOUD) corpus and CMU Multi-modal Opinion-level Sentiment Intensity (CMU-MOSI) corpus for multimodal sentiment analysis. Experimental results prove the effectiveness of the proposed approach. The accuracy of the MOUD and MOSI datasets is 90.43% and 82.71%, respectively. Compared to the state-of-the-art models, the improvement of the performance are approximately 2 and 0.4 points.","PeriodicalId":414793,"journal":{"name":"Proceedings of the 4th International Conference on Machine Learning and Soft Computing","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134285643","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：