Pub Date : 2015-03-03DOI: 10.1109/AISP.2015.7123484
Milad Vadoodparast, F. Taghiyareh
It is an interesting problem in a social system investigating how to affect a large number of people by just investing on a minority of them. This problem, i.e., influence maximization, is called “maximizing product adoption” in marketing applications. In this paper, we first propose a multi-agent framework called MAFIM to be used for maximizing product adoption in dynamic social networks. MAFIM consists of two types of agents: modeling agents and solution provider agents. These agents view a dynamic social network as consecutive static network snapshots and regarding that, choose a budget assignment policy so that each snapshot obtains its share from the budget defined by the sales manager. Based on MAFIM, we present MASPEL, a single product model which takes network communities, their judgments on each other and their profitabilities into account. MASPEL makes use of a specific budget assignment policy in which budgets are assigned to advertisement campaigns in a progressively decreasing manner. We applied our model on several real and synthetic dynamic social networks then evaluated the effectiveness of different campaign lengths. Our results show that it is more effective to launch many short-lived campaigns instead of few long-lived ones. It was also observed that betweenness has the best performance among centrality-based heuristics in leading the majority towards liking the advertised product.
{"title":"A multi-agent solution to maximizing product adoption in dynamic social networks","authors":"Milad Vadoodparast, F. Taghiyareh","doi":"10.1109/AISP.2015.7123484","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123484","url":null,"abstract":"It is an interesting problem in a social system investigating how to affect a large number of people by just investing on a minority of them. This problem, i.e., influence maximization, is called “maximizing product adoption” in marketing applications. In this paper, we first propose a multi-agent framework called MAFIM to be used for maximizing product adoption in dynamic social networks. MAFIM consists of two types of agents: modeling agents and solution provider agents. These agents view a dynamic social network as consecutive static network snapshots and regarding that, choose a budget assignment policy so that each snapshot obtains its share from the budget defined by the sales manager. Based on MAFIM, we present MASPEL, a single product model which takes network communities, their judgments on each other and their profitabilities into account. MASPEL makes use of a specific budget assignment policy in which budgets are assigned to advertisement campaigns in a progressively decreasing manner. We applied our model on several real and synthetic dynamic social networks then evaluated the effectiveness of different campaign lengths. Our results show that it is more effective to launch many short-lived campaigns instead of few long-lived ones. It was also observed that betweenness has the best performance among centrality-based heuristics in leading the majority towards liking the advertised product.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115111375","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 : 2015-03-03DOI: 10.1109/AISP.2015.7123506
Ahmadreza Ghaffarizdeh, M. Eftekhari, Donya Yazdani, Kamilia Ahmadi
In this paper, new ideas are presented for resolving the issues of two past self-organized criticality (SOC) evolutionary algorithms (EAs). The concept of SOC was first developed for modeling Mass Extinction and implemented by means of Sand Pile model in EAs. These types of EAs are especially employed when the optimization problems are multimodal in which preserving the diversity of solutions is a crucial task. Therefore analyzing the problems of SOC based EAs is worthwhile for making a progress in the field of multimodal optimization. Consequently, after an exact inspection of past research studies, the major shortcomings of previously developed algorithms are addressed which are twofold: firstly, the lack of avalanches in early generations, and secondly, the number of avalanches occurred in a population is out of proportion in terms of population size. In order to resolve these problems, some solutions are proposed in this study. The impact of these modifications are examined and illustrated by means of several benchmark optimization problems extracted from past research studies. Modified algorithm is compared and contrasted against previously developed SOC based algorithms and classical Genetic Algorithm (CGA). Results apparently show the effectiveness of eliminating addressed deficiencies in terms of accuracy and escaping from local optima.
{"title":"A new algorithm for improving deficiencies of past self-organized criticality based extinction algorithms","authors":"Ahmadreza Ghaffarizdeh, M. Eftekhari, Donya Yazdani, Kamilia Ahmadi","doi":"10.1109/AISP.2015.7123506","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123506","url":null,"abstract":"In this paper, new ideas are presented for resolving the issues of two past self-organized criticality (SOC) evolutionary algorithms (EAs). The concept of SOC was first developed for modeling Mass Extinction and implemented by means of Sand Pile model in EAs. These types of EAs are especially employed when the optimization problems are multimodal in which preserving the diversity of solutions is a crucial task. Therefore analyzing the problems of SOC based EAs is worthwhile for making a progress in the field of multimodal optimization. Consequently, after an exact inspection of past research studies, the major shortcomings of previously developed algorithms are addressed which are twofold: firstly, the lack of avalanches in early generations, and secondly, the number of avalanches occurred in a population is out of proportion in terms of population size. In order to resolve these problems, some solutions are proposed in this study. The impact of these modifications are examined and illustrated by means of several benchmark optimization problems extracted from past research studies. Modified algorithm is compared and contrasted against previously developed SOC based algorithms and classical Genetic Algorithm (CGA). Results apparently show the effectiveness of eliminating addressed deficiencies in terms of accuracy and escaping from local optima.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115206760","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 : 2015-03-03DOI: 10.1109/AISP.2015.7123512
E. Moeinaddini, Roya Ghasemkhani
Copyright protection of digital media becomes a major problem in internet due to its nature which allows gross duplication, alteration or even stolen of data. Several watermarking methods are reported in recent years to protect copyright of digital images. A good watermarking algorithm must satisfy imperceptibility, robustness and high data hiding capacity but without compromising for any of them. In this paper a novel blind watermarking scheme in Hadamard transform domain for digital image is proposed. For embedding watermark bits we modify the DHT coefficient of two adjacent blocks in same position. Watermark is a string of characters. This algorithm offers advantages of simpler implementation, low computation cost and high resiliency under compression. The experimental results and performance analysis shows good robustness under JPEG image compression and other common image processing operations like cropping, rotation, noise addition and filtering.
{"title":"A novel image watermarking scheme using blocks coefficient in DHT domain","authors":"E. Moeinaddini, Roya Ghasemkhani","doi":"10.1109/AISP.2015.7123512","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123512","url":null,"abstract":"Copyright protection of digital media becomes a major problem in internet due to its nature which allows gross duplication, alteration or even stolen of data. Several watermarking methods are reported in recent years to protect copyright of digital images. A good watermarking algorithm must satisfy imperceptibility, robustness and high data hiding capacity but without compromising for any of them. In this paper a novel blind watermarking scheme in Hadamard transform domain for digital image is proposed. For embedding watermark bits we modify the DHT coefficient of two adjacent blocks in same position. Watermark is a string of characters. This algorithm offers advantages of simpler implementation, low computation cost and high resiliency under compression. The experimental results and performance analysis shows good robustness under JPEG image compression and other common image processing operations like cropping, rotation, noise addition and filtering.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114025252","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 : 2015-03-03DOI: 10.1109/AISP.2015.7123525
Z. Naraghi, M. Jamzad
With the growing presence of computers in everyday life, communication improvement between human and machines is inevitable. Talking faces are the faces whose movements are synchronized to speech. They have an effective role in many applications. Lip is the most important part of a talking face. The main goal of this project is implementing a natural and human-like lip movement synthesis system for the Farsi language. For this purpose, a comprehensive audio visual database called SFAVD1 was designed and used. After extracting the sufficient features and designing a parallel Hidden Markov Model, the speech driven lip movement sequence generator system for Farsi input speech was implemented. To remove discontinuities between lip frames produced by the system, a morphing algorithm was used. The proposed system is unique for Farsi, and the evaluations have shown its acceptable quality.
{"title":"Speech driven lips animation for the Farsi language","authors":"Z. Naraghi, M. Jamzad","doi":"10.1109/AISP.2015.7123525","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123525","url":null,"abstract":"With the growing presence of computers in everyday life, communication improvement between human and machines is inevitable. Talking faces are the faces whose movements are synchronized to speech. They have an effective role in many applications. Lip is the most important part of a talking face. The main goal of this project is implementing a natural and human-like lip movement synthesis system for the Farsi language. For this purpose, a comprehensive audio visual database called SFAVD1 was designed and used. After extracting the sufficient features and designing a parallel Hidden Markov Model, the speech driven lip movement sequence generator system for Farsi input speech was implemented. To remove discontinuities between lip frames produced by the system, a morphing algorithm was used. The proposed system is unique for Farsi, and the evaluations have shown its acceptable quality.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134350289","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 : 2015-03-03DOI: 10.1109/AISP.2015.7123495
Maryam Dashti, R. Safabakhsh, Mohammadreza Pourfard, M. Abdollahifard
Video inpainting methods has a large number of applications and some of these algorithms are specialized for specific applications such as logo removal. There are only a few general video inpainting algorithms most of which are very time-consuming. This problem makes these algorithms unsuitable for fast video inpainting. In this paper, a fast simple logo removal algorithm has been proposed which uses frames of each video shot for logo removal and removes logo from video after a few iterations. A more accurate non-casual version of our algorithm is also proposed which uses both the information of previous and next frames. The quality of the inpainted video is also comparable with well-known video inpainting algorithms.
{"title":"Video logo removal using iterative subsequent matching","authors":"Maryam Dashti, R. Safabakhsh, Mohammadreza Pourfard, M. Abdollahifard","doi":"10.1109/AISP.2015.7123495","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123495","url":null,"abstract":"Video inpainting methods has a large number of applications and some of these algorithms are specialized for specific applications such as logo removal. There are only a few general video inpainting algorithms most of which are very time-consuming. This problem makes these algorithms unsuitable for fast video inpainting. In this paper, a fast simple logo removal algorithm has been proposed which uses frames of each video shot for logo removal and removes logo from video after a few iterations. A more accurate non-casual version of our algorithm is also proposed which uses both the information of previous and next frames. The quality of the inpainted video is also comparable with well-known video inpainting algorithms.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133768069","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 : 2015-03-01DOI: 10.1109/AISP.2015.7123504
Atefeh Zafarian, Ali Rokni, Shahram Khadivi, Sonia Ghiasifard
The shortage of the annotated training data is still an important challenge to building many Natural Language Process (NLP) tasks such as Named Entity Recognition. NER requires a large amount of training data with a high degree of human supervision whereas there is not enough labeled data for every language. In this paper, we use an unlabeled bilingual corpora to extract useful features from transferring information from resource-rich language toward resource-poor language and by using these features and a small training data, make a NER supervised model. Then we utilize a graph-based semi-supervised learning method that trains a CRF-based supervised classifier using that labeled data and uses high-confidence predictions on the unlabeled data to expand the training set and improve efficiency of NER model with the new training set.
{"title":"Semi-supervised learning for named entity recognition using weakly labeled training data","authors":"Atefeh Zafarian, Ali Rokni, Shahram Khadivi, Sonia Ghiasifard","doi":"10.1109/AISP.2015.7123504","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123504","url":null,"abstract":"The shortage of the annotated training data is still an important challenge to building many Natural Language Process (NLP) tasks such as Named Entity Recognition. NER requires a large amount of training data with a high degree of human supervision whereas there is not enough labeled data for every language. In this paper, we use an unlabeled bilingual corpora to extract useful features from transferring information from resource-rich language toward resource-poor language and by using these features and a small training data, make a NER supervised model. Then we utilize a graph-based semi-supervised learning method that trains a CRF-based supervised classifier using that labeled data and uses high-confidence predictions on the unlabeled data to expand the training set and improve efficiency of NER model with the new training set.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127662989","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 : 2015-03-01DOI: 10.1109/AISP.2015.7123518
Sonia Ghiasifard, Shahram Khadivi, M. Asadpour, Atefeh Zafarian
Community detection in social networks is usually done based on the density of connections between groups of nodes. However, these links do not necessarily represent an actual friendship especially in online social networks. There are users with declared friendship connections but without actual communication and no common interests. Most of the works in this area can be divided into two groups: topology-based and topic-based. The former usually leads to communities each containing diverse topics, and the latter leads to communities each with a consistent topic but with diverse structure. In this paper, we measure the similarity between users using topic models to generate virtual links for users with common interests. Moreover, in order to reduce the effect of useless links between users, we weight the network by measuring similarity of users' topics, so we could generate conforming communities, which contain only one topic or a group of consistent topics. The test results on Enron email dataset have shown the superior performance of our proposed method in the task of community detection.
{"title":"Improving the quality of overlapping community detection through link addition based on topic similarity","authors":"Sonia Ghiasifard, Shahram Khadivi, M. Asadpour, Atefeh Zafarian","doi":"10.1109/AISP.2015.7123518","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123518","url":null,"abstract":"Community detection in social networks is usually done based on the density of connections between groups of nodes. However, these links do not necessarily represent an actual friendship especially in online social networks. There are users with declared friendship connections but without actual communication and no common interests. Most of the works in this area can be divided into two groups: topology-based and topic-based. The former usually leads to communities each containing diverse topics, and the latter leads to communities each with a consistent topic but with diverse structure. In this paper, we measure the similarity between users using topic models to generate virtual links for users with common interests. Moreover, in order to reduce the effect of useless links between users, we weight the network by measuring similarity of users' topics, so we could generate conforming communities, which contain only one topic or a group of consistent topics. The test results on Enron email dataset have shown the superior performance of our proposed method in the task of community detection.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2015-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127364885","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 : 2014-08-25DOI: 10.1109/AISP.2015.7123501
Rasoul Ramezanian, Mostafa Salehi, Matteo Magnani, D. Montesi
The ways in which an innovation (e.g., new behaviour, idea, technology, product) diffuses among people can determine its success or failure. In this paper, we address the problem of diffusion of innovations over multiplex social networks where the neighbours of a person belong to one or multiple networks (or layers) such as friends, families, or colleagues. To this end, we generalise one of the basic game-theoretic diffusion models, called networked coordination game, for multiplex networks. We present analytical results for this extended model and validate them through a simulation study, finding among other properties a lower bound for the success of an innovation. While simple and leading to intuitively understandable results, to the best of our knowledge this is the first extension of a game-theoretic innovation diffusion model for multiplex networks and as such it provides a basic framework to study more sophisticated innovation dynamics.
{"title":"Diffusion of innovations over multiplex social networks","authors":"Rasoul Ramezanian, Mostafa Salehi, Matteo Magnani, D. Montesi","doi":"10.1109/AISP.2015.7123501","DOIUrl":"https://doi.org/10.1109/AISP.2015.7123501","url":null,"abstract":"The ways in which an innovation (e.g., new behaviour, idea, technology, product) diffuses among people can determine its success or failure. In this paper, we address the problem of diffusion of innovations over multiplex social networks where the neighbours of a person belong to one or multiple networks (or layers) such as friends, families, or colleagues. To this end, we generalise one of the basic game-theoretic diffusion models, called networked coordination game, for multiplex networks. We present analytical results for this extended model and validate them through a simulation study, finding among other properties a lower bound for the success of an innovation. While simple and leading to intuitively understandable results, to the best of our knowledge this is the first extension of a game-theoretic innovation diffusion model for multiplex networks and as such it provides a basic framework to study more sophisticated innovation dynamics.","PeriodicalId":405857,"journal":{"name":"2015 The International Symposium on Artificial Intelligence and Signal Processing (AISP)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116981636","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}