Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541307
L. Punchoojit, Nuttanont Hongwarittorrn
Prior studies did not examine how menu efficiency was related to menu components: icons, menu patterns, and label. Moreover, the research has never investigated whether label font size has an influence on findability of menu item. This study examined whether label font sizes influenced menu item selection time, on different menu design variations. The ANOVA test indicated that there was a significant effect on menu selection time. The results did not suggest that the significance was from the influence of label font size. However, the study found that menu pattern and icon shape had stronger influence on menu selection time.
{"title":"The Influence of Label Font Size on Menu Item Selection for Smartphone","authors":"L. Punchoojit, Nuttanont Hongwarittorrn","doi":"10.1109/ICAICTA.2018.8541307","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541307","url":null,"abstract":"Prior studies did not examine how menu efficiency was related to menu components: icons, menu patterns, and label. Moreover, the research has never investigated whether label font size has an influence on findability of menu item. This study examined whether label font sizes influenced menu item selection time, on different menu design variations. The ANOVA test indicated that there was a significant effect on menu selection time. The results did not suggest that the significance was from the influence of label font size. However, the study found that menu pattern and icon shape had stronger influence on menu selection time.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133223946","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541321
Santipong Thaiprayoon, A. Kongthon, C. Haruechaiyasak
Nowadays, search engine is an important tool for enabling users to search for information on the Internet. One of the most important problems of searching is inaccurate typing due to typographical and cognitive errors. Typographical errors are normally resulting from typing mistakes from adjacent letters on a keyboard layout. Cognitive errors are due to the lack of user knowledge in query term spelling. To solve the problems, we designed and developed a new version of Thai query correction program called ThaiQCor 2.0 that can handle both typographical and cognitive errors. Our program consists of two main approaches, word approximation and soundex. Word approximation employs the approximate string retrieval technique including character edit distance calculation. This approach aims to solve the typographical errors. Soundex applies the grapheme-to-phoneme conversion and then performs string matching approximation by calculating the edit distance of weighted phonemes from phoneme sequences. The objective of this approach is to handle the cognitive errors. All candidate words from both approaches are ranked based on their scores and suggested to the user. The experimental results showed that ThaiQCor 2.0 achieves the accuracy of 97.11% and 89.76% for place names and person names, respectively.
{"title":"ThaiQCor 2.0: Thai Query Correction via Soundex and Word Approximation","authors":"Santipong Thaiprayoon, A. Kongthon, C. Haruechaiyasak","doi":"10.1109/ICAICTA.2018.8541321","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541321","url":null,"abstract":"Nowadays, search engine is an important tool for enabling users to search for information on the Internet. One of the most important problems of searching is inaccurate typing due to typographical and cognitive errors. Typographical errors are normally resulting from typing mistakes from adjacent letters on a keyboard layout. Cognitive errors are due to the lack of user knowledge in query term spelling. To solve the problems, we designed and developed a new version of Thai query correction program called ThaiQCor 2.0 that can handle both typographical and cognitive errors. Our program consists of two main approaches, word approximation and soundex. Word approximation employs the approximate string retrieval technique including character edit distance calculation. This approach aims to solve the typographical errors. Soundex applies the grapheme-to-phoneme conversion and then performs string matching approximation by calculating the edit distance of weighted phonemes from phoneme sequences. The objective of this approach is to handle the cognitive errors. All candidate words from both approaches are ranked based on their scores and suggested to the user. The experimental results showed that ThaiQCor 2.0 achieves the accuracy of 97.11% and 89.76% for place names and person names, respectively.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125924124","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541291
Mio Kobayashi, T. Yoshinaga
The associative memory model comprised of coupled Gaussian maps is proposed. The Gaussian map is a one-dimensional discrete-time dynamical system, which generates various phenomena including periodic and non-periodic points. The Gaussian associative memory has similar characteristics of both Hopfield and chaos neural associative memories, and it can change those modes by just changing the system parameters. When the Gaussian associative memory successively recalls the stored patterns in such manner as the chaotic associative memory, the Gaussian associative memory also recalls some pseudo patterns which were not actually stored into the memory. It was found that the pseudo patterns corresponded to the chaotic trajectories generated in the Gaussian associative memory. Therefore, by using the method of avoiding chaotic behavior, we could eliminate the generation of the pseudo patterns. In this paper, we introduce the dynamics of the Gaussian associative memory model and present the simulation results. In addition, the output patterns obtained by the Gaussian associative memory with/without the function of avoiding chaos are presented.
{"title":"Associative Memory by Using Coupled Gaussian Maps","authors":"Mio Kobayashi, T. Yoshinaga","doi":"10.1109/ICAICTA.2018.8541291","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541291","url":null,"abstract":"The associative memory model comprised of coupled Gaussian maps is proposed. The Gaussian map is a one-dimensional discrete-time dynamical system, which generates various phenomena including periodic and non-periodic points. The Gaussian associative memory has similar characteristics of both Hopfield and chaos neural associative memories, and it can change those modes by just changing the system parameters. When the Gaussian associative memory successively recalls the stored patterns in such manner as the chaotic associative memory, the Gaussian associative memory also recalls some pseudo patterns which were not actually stored into the memory. It was found that the pseudo patterns corresponded to the chaotic trajectories generated in the Gaussian associative memory. Therefore, by using the method of avoiding chaotic behavior, we could eliminate the generation of the pseudo patterns. In this paper, we introduce the dynamics of the Gaussian associative memory model and present the simulation results. In addition, the output patterns obtained by the Gaussian associative memory with/without the function of avoiding chaos are presented.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114426194","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541310
Pisut Oncharoen, P. Vateekul
Recently, ability to handle tremendous amounts of information using increased computational capabilities has improved prediction of stock market behavior. Complex machine learning algorithms such as deep learning methods can analyze and detect complex data patterns. The recent prediction models use two types of inputs as (i) numerical information such as historical prices and technical indicators, and (ii) textual information including news contents or headlines. However, the use of textual data involves text representation construction. Traditional methods like word embedding may not be suitable for representing the semantics of financial news due to problems of word sparsity in datasets. In this paper, we aim to improve stock market predictions using a deep learning approach with event embedding vectors extracted from news headlines, historical price data, and a set of technical indicators as input. Our prediction model consists of Convolutional Neural Network (CNN) and Long Short-term Memory (LSTM) architectures. We use accuracy and annualized return based on trading simulation as performance metrics, and then perform experiments on three datasets obtained from different news sources namely Reuters, Reddit, and Intrinio. Results show that enhancing text representation vectors and considering both numerical and textual information as input to a deep neural network can improve prediction performance.
{"title":"Deep Learning for Stock Market Prediction Using Event Embedding and Technical Indicators","authors":"Pisut Oncharoen, P. Vateekul","doi":"10.1109/ICAICTA.2018.8541310","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541310","url":null,"abstract":"Recently, ability to handle tremendous amounts of information using increased computational capabilities has improved prediction of stock market behavior. Complex machine learning algorithms such as deep learning methods can analyze and detect complex data patterns. The recent prediction models use two types of inputs as (i) numerical information such as historical prices and technical indicators, and (ii) textual information including news contents or headlines. However, the use of textual data involves text representation construction. Traditional methods like word embedding may not be suitable for representing the semantics of financial news due to problems of word sparsity in datasets. In this paper, we aim to improve stock market predictions using a deep learning approach with event embedding vectors extracted from news headlines, historical price data, and a set of technical indicators as input. Our prediction model consists of Convolutional Neural Network (CNN) and Long Short-term Memory (LSTM) architectures. We use accuracy and annualized return based on trading simulation as performance metrics, and then perform experiments on three datasets obtained from different news sources namely Reuters, Reddit, and Intrinio. Results show that enhancing text representation vectors and considering both numerical and textual information as input to a deep neural network can improve prediction performance.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"126 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122240035","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}
Pub Date : 2018-08-01DOI: 10.1109/icaicta.2018.8541281
B. Sirinaovakul
Provides an abstract of the tutorial presentation and may include a brief professional biography of the presenter. The complete presentation was not made available for publication as part of the conference proceedings.
提供教程演示的摘要，可能包括演示者的简短专业简介。完整的报告没有作为会议记录的一部分提供出版。
{"title":"ICAICTA 2018 Tutorial","authors":"B. Sirinaovakul","doi":"10.1109/icaicta.2018.8541281","DOIUrl":"https://doi.org/10.1109/icaicta.2018.8541281","url":null,"abstract":"Provides an abstract of the tutorial presentation and may include a brief professional biography of the presenter. The complete presentation was not made available for publication as part of the conference proceedings.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"2016 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121325685","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541308
Pirapong Limprasitwong, C. Thongchaisuratkrul
This research aims to study an effective way of light using for plant growth. The light types included LED, grow and natural light. Investigated periods are germination and growth. A plant nursery was 1.2x1.2x1.5 m in dimension. The structure was made of PVC tube. It was covered by black canvas. The system was controlled by microcontroller. Two sensors modules DHT22 detected both of temperature and humidity. The plant nursery was separated into two rooms for LED and grow light testing. Cooling pads and water dispenser were used for the cooling system. A fan was installed for flowing air. The plant was watered automatically. From experimental result, the plant under LED light had the fastest rate of germination. The followed by grow light and the natural light respectively. The plant under grow light is the rapidest growth. The next are LED light and natural light, respectively.
{"title":"Plant Growth Using Automatic Control System under LED, Grow, and Natural Light","authors":"Pirapong Limprasitwong, C. Thongchaisuratkrul","doi":"10.1109/ICAICTA.2018.8541308","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541308","url":null,"abstract":"This research aims to study an effective way of light using for plant growth. The light types included LED, grow and natural light. Investigated periods are germination and growth. A plant nursery was 1.2x1.2x1.5 m in dimension. The structure was made of PVC tube. It was covered by black canvas. The system was controlled by microcontroller. Two sensors modules DHT22 detected both of temperature and humidity. The plant nursery was separated into two rooms for LED and grow light testing. Cooling pads and water dispenser were used for the cooling system. A fan was installed for flowing air. The plant was watered automatically. From experimental result, the plant under LED light had the fastest rate of germination. The followed by grow light and the natural light respectively. The plant under grow light is the rapidest growth. The next are LED light and natural light, respectively.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"88 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126201144","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541345
A. Dewandaru, S. Supriana, Saiful Akbar
The abundance of online news texts which contain embedded geographical name references from the internet provide motivation to produce higher level analysis in the form of thematic maps. This can be done by a performing automated geospatial information extraction and retrieval from relevant event-oriented corpora which mainly existed in natural language form. However, unified methods and framework available to address this transformation is still lacking. We propose the incorporation of unsupervised topic modeling and word embedding to help accomplishing the task of aggregating georeferenced data. The topic modeling tool would help suggesting the positive keywords and negative keywords for particular topic while the word embedding helped improve the recall score by extending the semanticaly similar keywords. The method was tested on Indonesian news corpus and achieved comparable result on two offical binary thematic maps case studies based on flood event in Jakarta and diphteria disease in Indonesia.
{"title":"Event-Oriented Map Extraction From Web News Portal : Binary Map Case Study on Diphteria Outbreak and Flood in Jakarta","authors":"A. Dewandaru, S. Supriana, Saiful Akbar","doi":"10.1109/ICAICTA.2018.8541345","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541345","url":null,"abstract":"The abundance of online news texts which contain embedded geographical name references from the internet provide motivation to produce higher level analysis in the form of thematic maps. This can be done by a performing automated geospatial information extraction and retrieval from relevant event-oriented corpora which mainly existed in natural language form. However, unified methods and framework available to address this transformation is still lacking. We propose the incorporation of unsupervised topic modeling and word embedding to help accomplishing the task of aggregating georeferenced data. The topic modeling tool would help suggesting the positive keywords and negative keywords for particular topic while the word embedding helped improve the recall score by extending the semanticaly similar keywords. The method was tested on Indonesian news corpus and achieved comparable result on two offical binary thematic maps case studies based on flood event in Jakarta and diphteria disease in Indonesia.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130037634","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541312
Wang Fengjiao, Masaki Aono
Nowadays, more and more people are getting used to social media such as Instagram, Facebook, Twitter, and Flickr to post images and texts to express their sentiment and emotions on almost all events and subjects. In consequence, analyzing sentiment of the huge number of images and texts on social networks has become more indispensable. Most of current research has focused on analyzing sentiment of textual data, while only few research has focused on sentiment analysis of image data. Some of these research has considered handcraft image features, the others has utilized Convolutional Neural Network (CNN) features. However, no research to our knowledge has considered mixing both hand-craft and CNN features. In this paper, we attempt to merge CNN which has shown remarkable achievements in Computer Vision recently, with handcraft features such as Color Histogram (CH) and Bag-of-Visual Words (BoVW) with some local features such as SURF and SIFT to predict sentiment of images. Furthermore, because it is often the case that the large amount of training data may not be easily obtained in the area of visual sentiment, we employ both data augmentation and transfer learning from a pre-trained CNN such as VGG16 trained with ImageNet dataset. With the handshake of hand-craft and End-to-End features from CNN, we attempt to attain the improvement of the performance of the proposed visual sentiment prediction framework. We conducted experiments on an image dataset from Twitter with polarity labels ("positive" and "negative"). The results of experiments demonstrate that our proposed visual sentimental prediction framework outperforms the current state-of-the-art methods.
{"title":"Visual Sentiment Prediction by Merging Hand-Craft and CNN Features","authors":"Wang Fengjiao, Masaki Aono","doi":"10.1109/ICAICTA.2018.8541312","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541312","url":null,"abstract":"Nowadays, more and more people are getting used to social media such as Instagram, Facebook, Twitter, and Flickr to post images and texts to express their sentiment and emotions on almost all events and subjects. In consequence, analyzing sentiment of the huge number of images and texts on social networks has become more indispensable. Most of current research has focused on analyzing sentiment of textual data, while only few research has focused on sentiment analysis of image data. Some of these research has considered handcraft image features, the others has utilized Convolutional Neural Network (CNN) features. However, no research to our knowledge has considered mixing both hand-craft and CNN features. In this paper, we attempt to merge CNN which has shown remarkable achievements in Computer Vision recently, with handcraft features such as Color Histogram (CH) and Bag-of-Visual Words (BoVW) with some local features such as SURF and SIFT to predict sentiment of images. Furthermore, because it is often the case that the large amount of training data may not be easily obtained in the area of visual sentiment, we employ both data augmentation and transfer learning from a pre-trained CNN such as VGG16 trained with ImageNet dataset. With the handshake of hand-craft and End-to-End features from CNN, we attempt to attain the improvement of the performance of the proposed visual sentiment prediction framework. We conducted experiments on an image dataset from Twitter with polarity labels (\"positive\" and \"negative\"). The results of experiments demonstrate that our proposed visual sentimental prediction framework outperforms the current state-of-the-art methods.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131517280","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541297
R. Rajagukguk, Masayu Leylia Khodra
We develop iSTS (Interpretable Semantic Textual Similarity) model to Indonesian corpus. System of iSTS is not only to represent the STS (Semantic Textual Similarity) score but also to give an explanation of the semantic similarity of the pair of sentence. The term of explanation refers to a pair of chunks with type such as EQUI, OPPO, SPE1, SPE2, REL, SIMI, NOALI and score ranged 0 to 5. Nowadays, iSTS corpus has not existed in the Indonesian version yet, by that mean we build that corpus. We adapt two best iSTS techniques for English corpus: VRep and UWB. VRep uses WordNet to representing word semantic, while UWB uses word embedding. Both of the techniques use similar process, such as preprocess, feature extraction, and classification. The adaptation of VRep and UWB on this research is performed by changing English resources in Indonesia such as WordNet, word embedding, etc. We also use four classifier as well as decision tree, SVM, random forest, and multilayer perceptron. VRep becomes the best model on type aspect and score aspect, while UWB becomes the best model on type + score aspect.
本文针对印尼语语料库建立了可解释语义文本相似度模型。语义文本相似度系统不仅表示语义文本相似度分数，而且对句子对的语义相似度给出解释。解释性术语是指EQUI、OPPO、SPE1、SPE2、REL、SIMI、NOALI等类型，得分范围为0 ~ 5的一对块。目前，ist的语料库还没有印尼语版本，因此我们建立了这个语料库。我们采用了两种最好的英语语料库列表技术:VRep和UWB。VRep使用WordNet来表示词语义，而UWB使用词嵌入。这两种技术使用类似的过程，如预处理、特征提取和分类。VRep和UWB在本研究中的适配是通过改变印度尼西亚的英语资源，如WordNet, word embedding等来实现的。我们还使用了四分类器以及决策树、支持向量机、随机森林和多层感知器。VRep在类型和分数方面成为最佳模型，UWB在类型+分数方面成为最佳模型。
{"title":"Interpretable Semantic Textual Similarity for Indonesian Sentence","authors":"R. Rajagukguk, Masayu Leylia Khodra","doi":"10.1109/ICAICTA.2018.8541297","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541297","url":null,"abstract":"We develop iSTS (Interpretable Semantic Textual Similarity) model to Indonesian corpus. System of iSTS is not only to represent the STS (Semantic Textual Similarity) score but also to give an explanation of the semantic similarity of the pair of sentence. The term of explanation refers to a pair of chunks with type such as EQUI, OPPO, SPE1, SPE2, REL, SIMI, NOALI and score ranged 0 to 5. Nowadays, iSTS corpus has not existed in the Indonesian version yet, by that mean we build that corpus. We adapt two best iSTS techniques for English corpus: VRep and UWB. VRep uses WordNet to representing word semantic, while UWB uses word embedding. Both of the techniques use similar process, such as preprocess, feature extraction, and classification. The adaptation of VRep and UWB on this research is performed by changing English resources in Indonesia such as WordNet, word embedding, etc. We also use four classifier as well as decision tree, SVM, random forest, and multilayer perceptron. VRep becomes the best model on type aspect and score aspect, while UWB becomes the best model on type + score aspect.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121123062","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}
Pub Date : 2018-08-01DOI: 10.1109/ICAICTA.2018.8541328
Tomonorio Kawamura, A. Kai, S. Nakagawa
The fundamental frequency (F0) is a quantity representing the pitch of periodic signal and its estimation for time-variant quasiperiodic acoustic signal is one of common problems in speech processing studies. The correct estimation of this contributes to the improvement of speech processing systems such as, analysis of prosody, test-to-speech system and speech recognition system. While many algorithms have been proposed and they exhibit excellent performance for clean environment, it is a very difficult task for noisy environment. It is generally known that machine learning approach is effective as a discriminative model for handling data in which noise is mixed. In this paper, we propose a robust fundamental frequency estimation method for noisy speech signal by using convolutional neural network (CNN) which is a of deep neural network (DNN). In our proposed method, convolution layer and pooling layer serve as an approximator of autocorrelation analysis and followed by discriminative modeling for classifying quantized F0 state. This process acquires a discriminator that extracts noise robust F0 features. Experimental result showed that our method outperforms convolutional methods based on autocorrelation analysis and its combination with DNN modeling.
{"title":"Noise Robust Fundamental Frequency Estimation of Speech using CNN-based discriminative modeling","authors":"Tomonorio Kawamura, A. Kai, S. Nakagawa","doi":"10.1109/ICAICTA.2018.8541328","DOIUrl":"https://doi.org/10.1109/ICAICTA.2018.8541328","url":null,"abstract":"The fundamental frequency (F0) is a quantity representing the pitch of periodic signal and its estimation for time-variant quasiperiodic acoustic signal is one of common problems in speech processing studies. The correct estimation of this contributes to the improvement of speech processing systems such as, analysis of prosody, test-to-speech system and speech recognition system. While many algorithms have been proposed and they exhibit excellent performance for clean environment, it is a very difficult task for noisy environment. It is generally known that machine learning approach is effective as a discriminative model for handling data in which noise is mixed. In this paper, we propose a robust fundamental frequency estimation method for noisy speech signal by using convolutional neural network (CNN) which is a of deep neural network (DNN). In our proposed method, convolution layer and pooling layer serve as an approximator of autocorrelation analysis and followed by discriminative modeling for classifying quantized F0 state. This process acquires a discriminator that extracts noise robust F0 features. Experimental result showed that our method outperforms convolutional methods based on autocorrelation analysis and its combination with DNN modeling.","PeriodicalId":184882,"journal":{"name":"2018 5th International Conference on Advanced Informatics: Concept Theory and Applications (ICAICTA)","volume":"354 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126685832","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：