Pub Date : 2014-12-15DOI: 10.1109/SPAC.2014.6982724
Cong Guo, Xinmei Tian
Social networks have been an important part of human's life. Online photo sharing websites like Flickr allow users to experience others' lifestyles by browsing photos. To gather users who have the same interests, the websites allow users to build their own interest groups and invite other users to join in. A commonly adopted recommendation in social networks such as Sina Microblog uses the social information of users. However, it performs poorly for inactive users. In this paper, we propose a group recommendation scheme by using both the content interest and social information of users. We use tag information, which is not only from users' photos but also from their favorite photos, to study the content interests of users and use the user-based collaborative filtering for recommendation. The trust-aware collaborative filtering is adopted to study the social information of users for recommendation. Finally, we combine the user-based collaborative filtering and trust-aware collaborative filtering to obtain a promising result on a real-world Flickr dataset.
{"title":"Flickr group recommendation using content interest and social information","authors":"Cong Guo, Xinmei Tian","doi":"10.1109/SPAC.2014.6982724","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982724","url":null,"abstract":"Social networks have been an important part of human's life. Online photo sharing websites like Flickr allow users to experience others' lifestyles by browsing photos. To gather users who have the same interests, the websites allow users to build their own interest groups and invite other users to join in. A commonly adopted recommendation in social networks such as Sina Microblog uses the social information of users. However, it performs poorly for inactive users. In this paper, we propose a group recommendation scheme by using both the content interest and social information of users. We use tag information, which is not only from users' photos but also from their favorite photos, to study the content interests of users and use the user-based collaborative filtering for recommendation. The trust-aware collaborative filtering is adopted to study the social information of users for recommendation. Finally, we combine the user-based collaborative filtering and trust-aware collaborative filtering to obtain a promising result on a real-world Flickr dataset.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126429227","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 : 2014-12-15DOI: 10.1109/SPAC.2014.6982683
Fudong Liu, Naiyang Guan, Yuhua Tang
Nonnegative matrix factorization (NMF) has been widely used to reduce dimensionality of data in image processing and various applications. Incorporating the geometric structure into NMF, graph regularized nonnegative matrix factorization (GNMF) has shown significant performance improvement in comparison to conventional NMF. However, both NMF and GNMF require the data matrix to reside in the memory, which gives rise to tremendous pressure for computation and storage. Moreover, this problem becomes serious if the datasets increase dramatically. In this paper, we propose an online GNMF (OGNMF) algorithm to process the incoming data in an incremental manner, i.e., OGNMF processes one data point or one chunk of data points one by one. By utilizing a smart buffering technique, OGNMF scales gracefully to large-scale datasets. Experimental results on text corpora demonstrate that OGNMF achieves better performance than the existing online NMF algorithms in terms of both accuracy and normalized mutual information, and outperforms the existing batch GNMF algorithms in terms of time overhead.
{"title":"Online graph regularized non-negative matrix factorization for streamming data","authors":"Fudong Liu, Naiyang Guan, Yuhua Tang","doi":"10.1109/SPAC.2014.6982683","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982683","url":null,"abstract":"Nonnegative matrix factorization (NMF) has been widely used to reduce dimensionality of data in image processing and various applications. Incorporating the geometric structure into NMF, graph regularized nonnegative matrix factorization (GNMF) has shown significant performance improvement in comparison to conventional NMF. However, both NMF and GNMF require the data matrix to reside in the memory, which gives rise to tremendous pressure for computation and storage. Moreover, this problem becomes serious if the datasets increase dramatically. In this paper, we propose an online GNMF (OGNMF) algorithm to process the incoming data in an incremental manner, i.e., OGNMF processes one data point or one chunk of data points one by one. By utilizing a smart buffering technique, OGNMF scales gracefully to large-scale datasets. Experimental results on text corpora demonstrate that OGNMF achieves better performance than the existing online NMF algorithms in terms of both accuracy and normalized mutual information, and outperforms the existing batch GNMF algorithms in terms of time overhead.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127240912","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 : 2014-12-15DOI: 10.1109/SPAC.2014.6982682
L. Gai
Collaborative filtering (CF) has been widely applied to improve the performance of recommendation systems. With the motivation of the Netflix Prize, researchers have proposed a series of CF algorithms for rating datasets, such as the 1 to 5 rating on Netflix. In this paper, we investigate the problem about implicit user feedback, which is a more common scenario (e.g. purchase history, click-through log, and page visitation). In these problems, the training data are only binary, reflecting the user's action or inaction. Under these circumstances, generating a personalized ranking list for every user is a more challenging task since we have less prior information. We consider it as a ranking problem: collaborative ranking (CR) skips the intermediate rating prediction step, and creates the ranked list directly. In order to solve the ranking problem, we propose a new model named pairwise probabilistic matrix factorization (PPMF), which takes a pairwise ranking approach integrated with the popular probabilistic matrix factorization (PMF) model to learn the relative preference for items. Experiments on benchmark datasets show that our proposed PPMF model outperforms the state-of-the-art implicit feedback collaborative ranking models by using different evaluation metrics.
{"title":"Pairwise probabilistic matrix factorization for implicit feedback collaborative filtering","authors":"L. Gai","doi":"10.1109/SPAC.2014.6982682","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982682","url":null,"abstract":"Collaborative filtering (CF) has been widely applied to improve the performance of recommendation systems. With the motivation of the Netflix Prize, researchers have proposed a series of CF algorithms for rating datasets, such as the 1 to 5 rating on Netflix. In this paper, we investigate the problem about implicit user feedback, which is a more common scenario (e.g. purchase history, click-through log, and page visitation). In these problems, the training data are only binary, reflecting the user's action or inaction. Under these circumstances, generating a personalized ranking list for every user is a more challenging task since we have less prior information. We consider it as a ranking problem: collaborative ranking (CR) skips the intermediate rating prediction step, and creates the ranked list directly. In order to solve the ranking problem, we propose a new model named pairwise probabilistic matrix factorization (PPMF), which takes a pairwise ranking approach integrated with the popular probabilistic matrix factorization (PMF) model to learn the relative preference for items. Experiments on benchmark datasets show that our proposed PPMF model outperforms the state-of-the-art implicit feedback collaborative ranking models by using different evaluation metrics.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121651282","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 : 2014-12-15DOI: 10.1109/SPAC.2014.6982706
Jinxiang Zhang, Jian Zhang
This paper proposes a novel face warping approach to deforming a 3D face using a few 3D facial features. The objective function includes two parts, the first part is local geometric constraints which aim to maintain the general characteristics of human face, and the second part is a shape regularization term which can force the deformed face to have some personalized facial characteristics. The problem can be solved by a least square optimization method. Experimental results demonstrate the effect of the proposed approach. And compared with existing method, the approach can generate better face warping result.
{"title":"Face warping based on local geometric constraint","authors":"Jinxiang Zhang, Jian Zhang","doi":"10.1109/SPAC.2014.6982706","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982706","url":null,"abstract":"This paper proposes a novel face warping approach to deforming a 3D face using a few 3D facial features. The objective function includes two parts, the first part is local geometric constraints which aim to maintain the general characteristics of human face, and the second part is a shape regularization term which can force the deformed face to have some personalized facial characteristics. The problem can be solved by a least square optimization method. Experimental results demonstrate the effect of the proposed approach. And compared with existing method, the approach can generate better face warping result.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134549270","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 : 2014-10-01DOI: 10.1109/SPAC.2014.6982700
J. Guo, Yanqing Guo, Lingyun Li, Ming Li
In recent years, plenty of advanced approaches for universal JPEG image steganalysis have been proposed due to the need of commercial and national security. Recently, a novel sparse-representation-based method was proposed, which applied sparse coding to image steganalysis [4]. Despite satisfying experimental results, the method emphasized too much on the role of l1-norm sparsity, while the effort of collaborative representation was totally ignored. In this paper, we focus on the least square problem in a binary classification model and present a similar yet much more efficient JPEG image steganalysis method based on collaborative representation. We still represent a testing sample collaboratively over the training samples from both classes (cover and stego), while the regularization term is changed from l1-norm to l2-norm and each class-specific representation residual owns an extra divisor. Experimental results show that our proposed steganalysis method performs better than the recently presented sparse-representation-based method as well as the traditional SVM-based method. Extensive experiments clearly show that our method has very competitive steganalysis performance, while it has significantly less complexity.
{"title":"A universal JPEG image steganalysis method based on collaborative representation","authors":"J. Guo, Yanqing Guo, Lingyun Li, Ming Li","doi":"10.1109/SPAC.2014.6982700","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982700","url":null,"abstract":"In recent years, plenty of advanced approaches for universal JPEG image steganalysis have been proposed due to the need of commercial and national security. Recently, a novel sparse-representation-based method was proposed, which applied sparse coding to image steganalysis [4]. Despite satisfying experimental results, the method emphasized too much on the role of l1-norm sparsity, while the effort of collaborative representation was totally ignored. In this paper, we focus on the least square problem in a binary classification model and present a similar yet much more efficient JPEG image steganalysis method based on collaborative representation. We still represent a testing sample collaboratively over the training samples from both classes (cover and stego), while the regularization term is changed from l1-norm to l2-norm and each class-specific representation residual owns an extra divisor. Experimental results show that our proposed steganalysis method performs better than the recently presented sparse-representation-based method as well as the traditional SVM-based method. Extensive experiments clearly show that our method has very competitive steganalysis performance, while it has significantly less complexity.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123370693","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 : 2014-10-01DOI: 10.1109/SPAC.2014.6982732
Chunzhi Wang, Dongyang Yu, Hui Xu, Hongwe Chen
A Peer-to-Peer (P2P) traffic identification method based on Bayesian trust sampling is presented in this paper, which predicts the fluctuation degree for next cycle of P2P traffic ratio, and optimizes for the used amount of historical proportion estimation. Simulation results show that, under the premise of using a fixed number of the estimated values for historical P2P ratio, this trust method makes a better forecast for the fluctuation degree of P2P traffic ratio, and reduces the amount of redundant samples.
{"title":"A Bayesian trust sampling method for P2P traffic inspection","authors":"Chunzhi Wang, Dongyang Yu, Hui Xu, Hongwe Chen","doi":"10.1109/SPAC.2014.6982732","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982732","url":null,"abstract":"A Peer-to-Peer (P2P) traffic identification method based on Bayesian trust sampling is presented in this paper, which predicts the fluctuation degree for next cycle of P2P traffic ratio, and optimizes for the used amount of historical proportion estimation. Simulation results show that, under the premise of using a fixed number of the estimated values for historical P2P ratio, this trust method makes a better forecast for the fluctuation degree of P2P traffic ratio, and reduces the amount of redundant samples.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121713706","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 : 2014-10-01DOI: 10.1109/SPAC.2014.6982713
Yi Mou, Xinge You, Xiubao Jiang, Duanquan Xu, Shujian Yu
The partial least squares (PLS) is designed for prediction problems when the number of predictors is larger than the number of training samples. PLS is based on latent components that are linear combinations of all of the original predictors, it automatically employs all predictors regardless of their relevance. This will degrade its performance and make it difficult to interpret the result. In this paper, global sparse PLS (GSPLS) is proposed to allow common variable selection in each deflation process as well as dimension reduction. We introduce the ℓ2, 1 norm to direction matrix and develop an algorithm for GSPLS via employing the Bregmen Iteration algorithm, illustrate the performance of proposed method with an analysis to red wine dataset. Numerical studies demonstrate the superiority of proposed GSPLS compared with standard PLS and other existing methods for variable selection and prediction in most of the cases.
{"title":"Global sparse partial least squares","authors":"Yi Mou, Xinge You, Xiubao Jiang, Duanquan Xu, Shujian Yu","doi":"10.1109/SPAC.2014.6982713","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982713","url":null,"abstract":"The partial least squares (PLS) is designed for prediction problems when the number of predictors is larger than the number of training samples. PLS is based on latent components that are linear combinations of all of the original predictors, it automatically employs all predictors regardless of their relevance. This will degrade its performance and make it difficult to interpret the result. In this paper, global sparse PLS (GSPLS) is proposed to allow common variable selection in each deflation process as well as dimension reduction. We introduce the ℓ2, 1 norm to direction matrix and develop an algorithm for GSPLS via employing the Bregmen Iteration algorithm, illustrate the performance of proposed method with an analysis to red wine dataset. Numerical studies demonstrate the superiority of proposed GSPLS compared with standard PLS and other existing methods for variable selection and prediction in most of the cases.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115487377","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 : 2014-10-01DOI: 10.1109/SPAC.2014.6982696
Wei Zhang, Xinge You
In this paper, we present a technique to describe local image features using scalable circumferential filters. Region description is the basic technique for many computer vision applications such as visual tracking, matching, and object recognition. In the first part of this paper, we present an elliptical cylinder projection method to geometrically normalize the ellipse regions to circular. In the second part of the paper, a set of scalable circumferential filters are proposed to extraction the distinctive feature of each region. Unlike traditional image filters, the shape of the proposed circumferential filters is fan and is scalable with its distance to the region's center. Experiments on typical images exhibit the robustness of the proposed method. Extensively quantitative evaluation and comparison demonstrate that the proposed method outperforms state-of-the-art method.
{"title":"Region description using scalable circumferential filters","authors":"Wei Zhang, Xinge You","doi":"10.1109/SPAC.2014.6982696","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982696","url":null,"abstract":"In this paper, we present a technique to describe local image features using scalable circumferential filters. Region description is the basic technique for many computer vision applications such as visual tracking, matching, and object recognition. In the first part of this paper, we present an elliptical cylinder projection method to geometrically normalize the ellipse regions to circular. In the second part of the paper, a set of scalable circumferential filters are proposed to extraction the distinctive feature of each region. Unlike traditional image filters, the shape of the proposed circumferential filters is fan and is scalable with its distance to the region's center. Experiments on typical images exhibit the robustness of the proposed method. Extensively quantitative evaluation and comparison demonstrate that the proposed method outperforms state-of-the-art method.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126777053","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 : 2014-10-01DOI: 10.1109/SPAC.2014.6982689
Shaodong Wang, Nan Wang, D. Tao, Lefei Zhang, Bo Du
Hyperspectral unmixing is a hot topic in signal and image processing. A high-dimensional data can be decomposed into two non-negative low-dimensional matrices by Non-negative matrix factorization(NMF). However, the algorithm has many local solutions because of the non-convexity of the objective function. Some algorithms solve this problem by adding auxiliary constraints, such as sparse. The sparse NMF has good performance but the result is unstable and sensitive to noise. Using the structural information for the unmixing approaches can make the decomposition stable. Someone used a clustering based on Euclidean distance to guide the decomposition and obtain good performance. The Euclidean distance is just used to measure the straight line distance of two points, and the ground objects usually obey certain statistical distribution. It's difficult to measure the difference between the statistical distributions comprehensively by Euclidean distance. KL divergence is a better metric. In this paper, we propose a new approach named KL divergence constrained NMF which measures the statistical distribution difference using KL divergence instead of the Euclidean distance. It can improve the accuracy of structured information by using the KL divergence in the algorithm. Experimental results based on synthetic and real hyperspectral data show the superiority of the proposed algorithm with respect to other state-of-the-art algorithms.
{"title":"A K-L divergence constrained sparse NMF for hyperspectral signal unmixing","authors":"Shaodong Wang, Nan Wang, D. Tao, Lefei Zhang, Bo Du","doi":"10.1109/SPAC.2014.6982689","DOIUrl":"https://doi.org/10.1109/SPAC.2014.6982689","url":null,"abstract":"Hyperspectral unmixing is a hot topic in signal and image processing. A high-dimensional data can be decomposed into two non-negative low-dimensional matrices by Non-negative matrix factorization(NMF). However, the algorithm has many local solutions because of the non-convexity of the objective function. Some algorithms solve this problem by adding auxiliary constraints, such as sparse. The sparse NMF has good performance but the result is unstable and sensitive to noise. Using the structural information for the unmixing approaches can make the decomposition stable. Someone used a clustering based on Euclidean distance to guide the decomposition and obtain good performance. The Euclidean distance is just used to measure the straight line distance of two points, and the ground objects usually obey certain statistical distribution. It's difficult to measure the difference between the statistical distributions comprehensively by Euclidean distance. KL divergence is a better metric. In this paper, we propose a new approach named KL divergence constrained NMF which measures the statistical distribution difference using KL divergence instead of the Euclidean distance. It can improve the accuracy of structured information by using the KL divergence in the algorithm. Experimental results based on synthetic and real hyperspectral data show the superiority of the proposed algorithm with respect to other state-of-the-art algorithms.","PeriodicalId":326246,"journal":{"name":"Proceedings 2014 IEEE International Conference on Security, Pattern Analysis, and Cybernetics (SPAC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2014-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134296051","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：