We consider the problem of adaptively routing a fleet of cooperative vehicles within a road network in the presence of uncertain and dynamic congestion conditions. To tackle this problem, we first propose a Gaussian Process Dynamic Congestion Model that can effectively characterize both the dynamics and the uncertainty of congestion conditions. Our model is efficient and thus facilitates real-time adaptive routing in the face of uncertainty. Using this congestion model, we develop an efficient algorithm for non-myopic adaptive routing to minimize the collective travel time of all vehicles in the system. A key property of our approach is the ability to efficiently reason about the long-term value of exploration, which enables collectively balancing the exploration/exploitation trade-off for entire fleets of vehicles. We validate our approach based on traffic data from two large Asian cities. We show that our congestion model is effective in modeling dynamic congestion conditions. We also show that our routing algorithm generates significantly faster routes compared to standard baselines, and achieves near-optimal performance compared to an omniscient routing algorithm. We also present the results from a preliminary field study, which showcases the efficacy of our approach.
{"title":"Adaptive collective routing using gaussian process dynamic congestion models","authors":"Siyuan Liu, Yisong Yue, R. Krishnan","doi":"10.1145/2487575.2487598","DOIUrl":"https://doi.org/10.1145/2487575.2487598","url":null,"abstract":"We consider the problem of adaptively routing a fleet of cooperative vehicles within a road network in the presence of uncertain and dynamic congestion conditions. To tackle this problem, we first propose a Gaussian Process Dynamic Congestion Model that can effectively characterize both the dynamics and the uncertainty of congestion conditions. Our model is efficient and thus facilitates real-time adaptive routing in the face of uncertainty. Using this congestion model, we develop an efficient algorithm for non-myopic adaptive routing to minimize the collective travel time of all vehicles in the system. A key property of our approach is the ability to efficiently reason about the long-term value of exploration, which enables collectively balancing the exploration/exploitation trade-off for entire fleets of vehicles. We validate our approach based on traffic data from two large Asian cities. We show that our congestion model is effective in modeling dynamic congestion conditions. We also show that our routing algorithm generates significantly faster routes compared to standard baselines, and achieves near-optimal performance compared to an omniscient routing algorithm. We also present the results from a preliminary field study, which showcases the efficacy of our approach.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"73 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82291052","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}
Pritam Gundecha, Suhas Ranganath, Zhuo Feng, Huan Liu
In recent years, social media sites have provided a large amount of information. Recipients of such information need mechanisms to know more about the received information, including the provenance. Previous research has shown that some attributes related to the received information provide additional context, so that a recipient can assess the amount of value, trust, and validity to be placed in the received information. Personal attributes of a user, including name, location, education, ethnicity, gender, and political and religious affiliations, can be found in social media sites. In this paper, we present a novel web-based tool for collecting the attributes of interest associated with a particular social media user related to the received information. This tool provides a way to combine different attributes available at different social media sites into a single user profile. Using different types of Twitter users, we also evaluate the performance of the tool in terms of number of attribute values collected, validity of these values, and total amount of retrieval time.
{"title":"A tool for collecting provenance data in social media","authors":"Pritam Gundecha, Suhas Ranganath, Zhuo Feng, Huan Liu","doi":"10.1145/2487575.2487713","DOIUrl":"https://doi.org/10.1145/2487575.2487713","url":null,"abstract":"In recent years, social media sites have provided a large amount of information. Recipients of such information need mechanisms to know more about the received information, including the provenance. Previous research has shown that some attributes related to the received information provide additional context, so that a recipient can assess the amount of value, trust, and validity to be placed in the received information. Personal attributes of a user, including name, location, education, ethnicity, gender, and political and religious affiliations, can be found in social media sites. In this paper, we present a novel web-based tool for collecting the attributes of interest associated with a particular social media user related to the received information. This tool provides a way to combine different attributes available at different social media sites into a single user profile. Using different types of Twitter users, we also evaluate the performance of the tool in terms of number of attribute values collected, validity of these values, and total amount of retrieval time.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86383568","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 problem of point of interest (POI) recommendation is to provide personalized recommendations of places of interests, such as restaurants, for mobile users. Due to its complexity and its connection to location based social networks (LBSNs), the decision process of a user choose a POI is complex and can be influenced by various factors, such as user preferences, geographical influences, and user mobility behaviors. While there are some studies on POI recommendations, it lacks of integrated analysis of the joint effect of multiple factors. To this end, in this paper, we propose a novel geographical probabilistic factor analysis framework which strategically takes various factors into consideration. Specifically, this framework allows to capture the geographical influences on a user's check-in behavior. Also, the user mobility behaviors can be effectively exploited in the recommendation model. Moreover, the recommendation model can effectively make use of user check-in count data as implicity user feedback for modeling user preferences. Finally, experimental results on real-world LBSNs data show that the proposed recommendation method outperforms state-of-the-art latent factor models with a significant margin.
{"title":"Learning geographical preferences for point-of-interest recommendation","authors":"B. Liu, Yanjie Fu, Zijun Yao, Hui Xiong","doi":"10.1145/2487575.2487673","DOIUrl":"https://doi.org/10.1145/2487575.2487673","url":null,"abstract":"The problem of point of interest (POI) recommendation is to provide personalized recommendations of places of interests, such as restaurants, for mobile users. Due to its complexity and its connection to location based social networks (LBSNs), the decision process of a user choose a POI is complex and can be influenced by various factors, such as user preferences, geographical influences, and user mobility behaviors. While there are some studies on POI recommendations, it lacks of integrated analysis of the joint effect of multiple factors. To this end, in this paper, we propose a novel geographical probabilistic factor analysis framework which strategically takes various factors into consideration. Specifically, this framework allows to capture the geographical influences on a user's check-in behavior. Also, the user mobility behaviors can be effectively exploited in the recommendation model. Moreover, the recommendation model can effectively make use of user check-in count data as implicity user feedback for modeling user preferences. Finally, experimental results on real-world LBSNs data show that the proposed recommendation method outperforms state-of-the-art latent factor models with a significant margin.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"45 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80550347","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}
Identifying the k most influential individuals in a social network is a well-studied problem. The objective is to detect k individuals in a (social) network who will influence the maximum number of people, if they are independently convinced of adopting a new strategy (product, idea, etc). There are cases in real life, however, where we aim to instigate groups instead of individuals to trigger network diffusion. Such cases abound, e.g., billboards, TV commercials and newspaper ads are utilized extensively to boost the popularity and raise awareness. In this paper, we generalize the "influential nodes" problem. Namely we are interested to locate the most "influential groups" in a network. As the first paper to address this problem: we (1) propose a fine-grained model of information diffusion for the group-based problem, (2) show that the process is submodular and present an algorithm to determine the influential groups under this model (with a precise approximation bound), (3) propose a coarse-grained model that inspects the network at group level (not individuals) significantly speeding up calculations for large networks, (4) show that the diffusion function we design here is submodular in general case, and propose an approximation algorithm for this coarse-grained model, and finally by conducting experiments on real datasets, (5) demonstrate that seeding members of selected groups to be the first adopters can broaden diffusion (when compared to the influential individuals case). Moreover, we can identify these influential groups much faster (up to 12 million times speedup), delivering a practical solution to this problem.
{"title":"Information cascade at group scale","authors":"Milad Eftekhar, Y. Ganjali, Nick Koudas","doi":"10.1145/2487575.2487683","DOIUrl":"https://doi.org/10.1145/2487575.2487683","url":null,"abstract":"Identifying the k most influential individuals in a social network is a well-studied problem. The objective is to detect k individuals in a (social) network who will influence the maximum number of people, if they are independently convinced of adopting a new strategy (product, idea, etc). There are cases in real life, however, where we aim to instigate groups instead of individuals to trigger network diffusion. Such cases abound, e.g., billboards, TV commercials and newspaper ads are utilized extensively to boost the popularity and raise awareness. In this paper, we generalize the \"influential nodes\" problem. Namely we are interested to locate the most \"influential groups\" in a network. As the first paper to address this problem: we (1) propose a fine-grained model of information diffusion for the group-based problem, (2) show that the process is submodular and present an algorithm to determine the influential groups under this model (with a precise approximation bound), (3) propose a coarse-grained model that inspects the network at group level (not individuals) significantly speeding up calculations for large networks, (4) show that the diffusion function we design here is submodular in general case, and propose an approximation algorithm for this coarse-grained model, and finally by conducting experiments on real datasets, (5) demonstrate that seeding members of selected groups to be the first adopters can broaden diffusion (when compared to the influential individuals case). Moreover, we can identify these influential groups much faster (up to 12 million times speedup), delivering a practical solution to this problem.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"93 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83826684","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}
Hongzhi Yin, Yizhou Sun, B. Cui, Zhiting Hu, Ling Chen
Newly emerging location-based and event-based social network services provide us with a new platform to understand users' preferences based on their activity history. A user can only visit a limited number of venues/events and most of them are within a limited distance range, so the user-item matrix is very sparse, which creates a big challenge for traditional collaborative filtering-based recommender systems. The problem becomes more challenging when people travel to a new city where they have no activity history. In this paper, we propose LCARS, a location-content-aware recommender system that offers a particular user a set of venues (e.g., restaurants) or events (e.g., concerts and exhibitions) by giving consideration to both personal interest and local preference. This recommender system can facilitate people's travel not only near the area in which they live, but also in a city that is new to them. Specifically, LCARS consists of two components: offline modeling and online recommendation. The offline modeling part, called LCA-LDA, is designed to learn the interest of each individual user and the local preference of each individual city by capturing item co-occurrence patterns and exploiting item contents. The online recommendation part automatically combines the learnt interest of the querying user and the local preference of the querying city to produce the top-k recommendations. To speed up this online process, a scalable query processing technique is developed by extending the classic Threshold Algorithm (TA). We evaluate the performance of our recommender system on two large-scale real data sets, DoubanEvent and Foursquare. The results show the superiority of LCARS in recommending spatial items for users, especially when traveling to new cities, in terms of both effectiveness and efficiency.
{"title":"LCARS: a location-content-aware recommender system","authors":"Hongzhi Yin, Yizhou Sun, B. Cui, Zhiting Hu, Ling Chen","doi":"10.1145/2487575.2487608","DOIUrl":"https://doi.org/10.1145/2487575.2487608","url":null,"abstract":"Newly emerging location-based and event-based social network services provide us with a new platform to understand users' preferences based on their activity history. A user can only visit a limited number of venues/events and most of them are within a limited distance range, so the user-item matrix is very sparse, which creates a big challenge for traditional collaborative filtering-based recommender systems. The problem becomes more challenging when people travel to a new city where they have no activity history. In this paper, we propose LCARS, a location-content-aware recommender system that offers a particular user a set of venues (e.g., restaurants) or events (e.g., concerts and exhibitions) by giving consideration to both personal interest and local preference. This recommender system can facilitate people's travel not only near the area in which they live, but also in a city that is new to them. Specifically, LCARS consists of two components: offline modeling and online recommendation. The offline modeling part, called LCA-LDA, is designed to learn the interest of each individual user and the local preference of each individual city by capturing item co-occurrence patterns and exploiting item contents. The online recommendation part automatically combines the learnt interest of the querying user and the local preference of the querying city to produce the top-k recommendations. To speed up this online process, a scalable query processing technique is developed by extending the classic Threshold Algorithm (TA). We evaluate the performance of our recommender system on two large-scale real data sets, DoubanEvent and Foursquare. The results show the superiority of LCARS in recommending spatial items for users, especially when traveling to new cities, in terms of both effectiveness and efficiency.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79055830","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}
Network science has emerged over the last years as an interdisciplinary area spanning traditional domains including mathematics, computer science, sociology, biology and economics. Since complexity in social, biological and economical systems, and more generally in complex systems, arises through pairwise interactions there exists a surging interest in understanding networks. In this tutorial, we will provide an in-depth presentation of the most popular random-graph models used for modeling real-world networks. We will then discuss efficient algorithmic techniques for mining large graphs, with emphasis on the problems of extracting graph sparsifiers, partitioning graphs into densely connected components, and finding dense subgraphs. We will motivate the problems we will discuss and the algorithms we will present with real-world applications. Our aim is to survey important results in the areas of modeling and mining large graphs, to uncover the intuition behind the key ideas, and to present future research directions.
{"title":"Algorithmic techniques for modeling and mining large graphs (AMAzING)","authors":"A. Frieze, A. Gionis, Charalampos E. Tsourakakis","doi":"10.1145/2487575.2506176","DOIUrl":"https://doi.org/10.1145/2487575.2506176","url":null,"abstract":"Network science has emerged over the last years as an interdisciplinary area spanning traditional domains including mathematics, computer science, sociology, biology and economics. Since complexity in social, biological and economical systems, and more generally in complex systems, arises through pairwise interactions there exists a surging interest in understanding networks. In this tutorial, we will provide an in-depth presentation of the most popular random-graph models used for modeling real-world networks. We will then discuss efficient algorithmic techniques for mining large graphs, with emphasis on the problems of extracting graph sparsifiers, partitioning graphs into densely connected components, and finding dense subgraphs. We will motivate the problems we will discuss and the algorithms we will present with real-world applications. Our aim is to survey important results in the areas of modeling and mining large graphs, to uncover the intuition behind the key ideas, and to present future research directions.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"82 4 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83444755","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 key algorithmic problem in viral marketing is to identify a set of influential users (called seeds) in a social network, who, when convinced to adopt a product, shall influence other users in the network, leading to a large number of adoptions. When two or more players compete with similar products on the same network we talk about competitive viral marketing, which so far has been studied exclusively from the perspective of one of the competing players. In this paper we propose and study the novel problem of competitive viral marketing from the perspective of the host, i.e., the owner of the social network platform. The host sells viral marketing campaigns as a service to its customers, keeping control of the selection of seeds. Each company specifies its budget and the host allocates the seeds accordingly. From the host's perspective, it is important not only to choose the seeds to maximize the collective expected spread, but also to assign seeds to companies so that it guarantees the "bang for the buck" for all companies is nearly identical, which we formalize as the fair seed allocation problem. We propose a new propagation model capturing the competitive nature of viral marketing. Our model is intuitive and retains the desired properties of monotonicity and submodularity. We show that the fair seed allocation problem is NP-hard, and develop an efficient algorithm called Needy Greedy. We run experiments on three real-world social networks, showing that our algorithm is effective and scalable.
{"title":"The bang for the buck: fair competitive viral marketing from the host perspective","authors":"Wei Lu, F. Bonchi, Amit Goyal, L. Lakshmanan","doi":"10.1145/2487575.2487649","DOIUrl":"https://doi.org/10.1145/2487575.2487649","url":null,"abstract":"The key algorithmic problem in viral marketing is to identify a set of influential users (called seeds) in a social network, who, when convinced to adopt a product, shall influence other users in the network, leading to a large number of adoptions. When two or more players compete with similar products on the same network we talk about competitive viral marketing, which so far has been studied exclusively from the perspective of one of the competing players. In this paper we propose and study the novel problem of competitive viral marketing from the perspective of the host, i.e., the owner of the social network platform. The host sells viral marketing campaigns as a service to its customers, keeping control of the selection of seeds. Each company specifies its budget and the host allocates the seeds accordingly. From the host's perspective, it is important not only to choose the seeds to maximize the collective expected spread, but also to assign seeds to companies so that it guarantees the \"bang for the buck\" for all companies is nearly identical, which we formalize as the fair seed allocation problem. We propose a new propagation model capturing the competitive nature of viral marketing. Our model is intuitive and retains the desired properties of monotonicity and submodularity. We show that the fair seed allocation problem is NP-hard, and develop an efficient algorithm called Needy Greedy. We run experiments on three real-world social networks, showing that our algorithm is effective and scalable.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"44 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81180903","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}
Fred Morstatter, Shamanth Kumar, Huan Liu, Ross Maciejewski
In the era of big data it is increasingly difficult for an analyst to extract meaningful knowledge from a sea of information. We present TweetXplorer, a system for analysts with little information about an event to gain knowledge through the use of effective visualization techniques. Using tweets collected during Hurricane Sandy as an example, we will lead the reader through a workflow that exhibits the functionality of the system.
{"title":"Understanding Twitter data with TweetXplorer","authors":"Fred Morstatter, Shamanth Kumar, Huan Liu, Ross Maciejewski","doi":"10.1145/2487575.2487703","DOIUrl":"https://doi.org/10.1145/2487575.2487703","url":null,"abstract":"In the era of big data it is increasingly difficult for an analyst to extract meaningful knowledge from a sea of information. We present TweetXplorer, a system for analysts with little information about an event to gain knowledge through the use of effective visualization techniques. Using tweets collected during Hurricane Sandy as an example, we will lead the reader through a workflow that exhibits the functionality of the system.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87235783","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}
Shuo Xiang, Lei Yuan, Wei Fan, Yalin Wang, P. Thompson, Jieping Ye
With the advances and increasing sophistication in data collection techniques, we are facing with large amounts of data collected from multiple heterogeneous sources in many applications. For example, in the study of Alzheimer's Disease (AD), different types of measurements such as neuroimages, gene/protein expression data, genetic data etc. are often collected and analyzed together for improved predictive power. It is believed that a joint learning of multiple data sources is beneficial as different data sources may contain complementary information, and feature-pruning and data source selection are critical for learning interpretable models from high-dimensional data. Very often the collected data comes with block-wise missing entries; for example, a patient without the MRI scan will have no information in the MRI data block, making his/her overall record incomplete. There has been a growing interest in the data mining community on expanding traditional techniques for single-source complete data analysis to the study of multi-source incomplete data. The key challenge is how to effectively integrate information from multiple heterogeneous sources in the presence of block-wise missing data. In this paper we first investigate the situation of complete data and present a unified ``bi-level" learning model for multi-source data. Then we give a natural extension of this model to the more challenging case with incomplete data. Our major contributions are threefold: (1) the proposed models handle both feature-level and source-level analysis in a unified formulation and include several existing feature learning approaches as special cases; (2) the model for incomplete data avoids direct imputation of the missing elements and thus provides superior performances. Moreover, it can be easily generalized to other applications with block-wise missing data sources; (3) efficient optimization algorithms are presented for both the complete and incomplete models. We have performed comprehensive evaluations of the proposed models on the application of AD diagnosis. Our proposed models compare favorably against existing approaches.
{"title":"Multi-source learning with block-wise missing data for Alzheimer's disease prediction","authors":"Shuo Xiang, Lei Yuan, Wei Fan, Yalin Wang, P. Thompson, Jieping Ye","doi":"10.1145/2487575.2487594","DOIUrl":"https://doi.org/10.1145/2487575.2487594","url":null,"abstract":"With the advances and increasing sophistication in data collection techniques, we are facing with large amounts of data collected from multiple heterogeneous sources in many applications. For example, in the study of Alzheimer's Disease (AD), different types of measurements such as neuroimages, gene/protein expression data, genetic data etc. are often collected and analyzed together for improved predictive power. It is believed that a joint learning of multiple data sources is beneficial as different data sources may contain complementary information, and feature-pruning and data source selection are critical for learning interpretable models from high-dimensional data. Very often the collected data comes with block-wise missing entries; for example, a patient without the MRI scan will have no information in the MRI data block, making his/her overall record incomplete. There has been a growing interest in the data mining community on expanding traditional techniques for single-source complete data analysis to the study of multi-source incomplete data. The key challenge is how to effectively integrate information from multiple heterogeneous sources in the presence of block-wise missing data. In this paper we first investigate the situation of complete data and present a unified ``bi-level\" learning model for multi-source data. Then we give a natural extension of this model to the more challenging case with incomplete data. Our major contributions are threefold: (1) the proposed models handle both feature-level and source-level analysis in a unified formulation and include several existing feature learning approaches as special cases; (2) the model for incomplete data avoids direct imputation of the missing elements and thus provides superior performances. Moreover, it can be easily generalized to other applications with block-wise missing data sources; (3) efficient optimization algorithms are presented for both the complete and incomplete models. We have performed comprehensive evaluations of the proposed models on the application of AD diagnosis. Our proposed models compare favorably against existing approaches.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"30 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90576930","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}
I. E. Yen, Chun-Fu Chang, Ting-Wei Lin, Shan-Wei Lin, Shou-De Lin
Linear Classification has achieved complexity linear to the data size. However, in many applications, data contain large amount of samples that does not help improve the quality of model, but still cost much I/O and memory to process. In this paper, we show how a Block Coordinate Descent method based on Nearest-Neighbor Index can significantly reduce such cost when learning a dual-sparse model. In particular, we employ truncated loss function to induce a series of convex programs with superior dual sparsity, and solve each dual using Indexed Block Coordinate Descent, which makes use of Approximate Nearest Neighbor (ANN) search to select active dual variables without I/O cost on irrelevant samples. We prove that, despite the bias and weak guarantee from ANN query, the proposed algorithm has global convergence to the solution defined on entire dataset, with sublinear complexity each iteration. Experiments in both sufficient and limited memory conditions show that the proposed approach learns many times faster than other state-of-the-art solvers without sacrificing accuracy.
{"title":"Indexed block coordinate descent for large-scale linear classification with limited memory","authors":"I. E. Yen, Chun-Fu Chang, Ting-Wei Lin, Shan-Wei Lin, Shou-De Lin","doi":"10.1145/2487575.2487626","DOIUrl":"https://doi.org/10.1145/2487575.2487626","url":null,"abstract":"Linear Classification has achieved complexity linear to the data size. However, in many applications, data contain large amount of samples that does not help improve the quality of model, but still cost much I/O and memory to process. In this paper, we show how a Block Coordinate Descent method based on Nearest-Neighbor Index can significantly reduce such cost when learning a dual-sparse model. In particular, we employ truncated loss function to induce a series of convex programs with superior dual sparsity, and solve each dual using Indexed Block Coordinate Descent, which makes use of Approximate Nearest Neighbor (ANN) search to select active dual variables without I/O cost on irrelevant samples. We prove that, despite the bias and weak guarantee from ANN query, the proposed algorithm has global convergence to the solution defined on entire dataset, with sublinear complexity each iteration. Experiments in both sufficient and limited memory conditions show that the proposed approach learns many times faster than other state-of-the-art solvers without sacrificing accuracy.","PeriodicalId":20472,"journal":{"name":"Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining","volume":"52 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2013-08-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86863601","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：