Results 1 -
8 of
8
Detecting and Tracking Coordinated Groups in Dense, Systematically Moving, Crowds
"... We address the problem of detecting and tracking clusters of moving objects in very noisy environments. Monitoring a crowded football stadium for small groups of individuals acting suspiciously is an example instance of this problem. In this example the vast majority of individuals are not part of a ..."
Abstract
-
Cited by 2 (0 self)
- Add to MetaCart
(Show Context)
We address the problem of detecting and tracking clusters of moving objects in very noisy environments. Monitoring a crowded football stadium for small groups of individuals acting suspiciously is an example instance of this problem. In this example the vast majority of individuals are not part of a suspicious group and are considered as noise. Existing spatio-temporal cluster algorithms are only capable of detecting small clusters in extreme noise when the noise objects are moving randomly. In reality, including the example cited, the noise objects move more systematically instead of moving randomly. The members of the suspicious groups attempt to mimic the behaviors of the crowd in order to blend in and avoid detection. This significantly exacerbates the problem of detecting the true clusters. We propose the use of Support Vector Machines (SVMs) to differentiate the true clusters and their members from the systematically moving noise objects. Our technique utilizes the relational history of the moving objects, implicitly tracked in a relationship graph, and a SVM to increase the accuracy of the clustering algorithm. A modified DBSCAN algorithm is then used to discover clusters of highly related objects from the relationship graph. We evaluate our technique experimentally on several data sets of mobile objects. The experiments show that our technique is able to accurately and efficiently identify groups of suspicious individuals in dense crowds. 1
Spatial Interestingness Measures for Co-location Pattern Mining
"... Abstract—Co-location pattern mining aims at finding sub-sets of spatial features frequently located together in spatial proximity. The underlying motivation is to model the spatial correlation structure between the features. This allows to discover interesting co-location rules (feature interactions ..."
Abstract
-
Cited by 1 (0 self)
- Add to MetaCart
(Show Context)
Abstract—Co-location pattern mining aims at finding sub-sets of spatial features frequently located together in spatial proximity. The underlying motivation is to model the spatial correlation structure between the features. This allows to discover interesting co-location rules (feature interactions) for spatial analysis and prediction tasks. As in association rule mining, a major problem is the huge amount of possible patterns and rules. Hence, measures are needed to identify interesting patterns and rules. Existing approaches so far focused on finding frequent patterns, patterns including rare features, and patterns occurring in small (local) regions. In this paper, we present a new general class of interest-ingness measures that are based on the spatial distribution of co-location patterns. These measures allow to judge the interestingness of a pattern based on properties of the un-derlying spatial feature distribution. The results are different from standard measures like participation index or confidence. To demonstrate the usefulness of these measures, we apply our approach to the discovery of rules on a subset of the OpenStreetMap point-of-interest data. Keywords-Co-location pattern mining, interestingness mea-sures, density estimation I.
Regional Co-locations of Arbitrary Shapes
"... Abstract. In many application domains, occurrences of related spatial features may exhibit co-location pattern. For example, some disease may be in spatial proximity of certain type of pollution. This paper studies the problem of regional co-locations with arbitrary shapes. Regional colocations repr ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract. In many application domains, occurrences of related spatial features may exhibit co-location pattern. For example, some disease may be in spatial proximity of certain type of pollution. This paper studies the problem of regional co-locations with arbitrary shapes. Regional colocations represent regions in which two spatial features exhibit stronger or weaker co-location than that in other regions. Finding regional colocations of arbitrary shapes is very challenging because: (1) statistical frameworks for mining regional co-location do not exist; and (2) testing all possible arbitrary shaped regions is computational prohibitive even for very small dataset. In this paper, we propose frequentist and Bayesian frameworks for mining regional co-locations and develop a probabilistic expansion heuristic to find arbitrary shaped regions. Experimental results on synthetic and real world data show that both frequentist method and Bayesian statistical approach can recover the region with arbitrary shapes. Our approaches outperform baseline algorithms in terms of F measure. Bayesian statistical approach is approximately three orders of magnitude faster than the frequentist approach. 1
Regional Pattern Discovery in Geo-Referenced Datasets Using PCA
"... Abstract. Existing data mining techniques mostly focus on finding global patterns and lack the ability to systematically discover regional patterns. Most relationships in spatial datasets are regional; therefore there is a great need to extract regional knowledge from spatial datasets. This paper pr ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract. Existing data mining techniques mostly focus on finding global patterns and lack the ability to systematically discover regional patterns. Most relationships in spatial datasets are regional; therefore there is a great need to extract regional knowledge from spatial datasets. This paper proposes a novel framework to discover interesting regions characterized by “strong regional correlation relationships ” between attributes, and methods to analyze differences and similarities between regions. The framework employs a twophase approach: it first discovers regions by employing clustering algorithms that maximize a PCA-based fitness function and then applies post processing techniques to explain underlying regional structures and correlation patterns. Additionally, a new similarity measure that assesses the structural similarity of regions based on correlation sets is introduced. We evaluate our framework in a case study which centers on finding correlations between arsenic pollution and other factors in water wells and demonstrate that our framework effectively identifies regional correlation patterns.
Incremental and Parallel Association Mining for Evolving Spatial Data: A Less Iterative Approach on MapReduce
"... Spatial association mining has been used for discovering frequent spatial relationship patterns from a large spatial database. When the database is constantly updated with fresh data, it is computationally expensive to redo the pat-tern discovery process for the updated database. This work explores ..."
Abstract
- Add to MetaCart
(Show Context)
Spatial association mining has been used for discovering frequent spatial relationship patterns from a large spatial database. When the database is constantly updated with fresh data, it is computationally expensive to redo the pat-tern discovery process for the updated database. This work explores methods for finding spatial association patterns in-crementally from evolving spatial databases. For large-scale spatial data processing, distributed and parallel computing platforms are becoming more popular. This paper presents a parallel and incremental algorithm for the spatial pat-tern update processing on the MapReduce framework. The experimental evaluation shows that our algorithmic design choice is effective for the MapReduce with reducing an it-erative processing and can achieve significant performance gains. 1
OF THE UNIVERSITY OF MINNESOTA BY
, 2012
"... I had the honor of working with a number of amazing individuals during my time at the University of Minnesota. This thesis is the outcome of that wonderful association. First, I express my gratitude to my advisor, Professor. Shekhar, for being such a wonderful mentor. I am particularly grateful for ..."
Abstract
- Add to MetaCart
(Show Context)
I had the honor of working with a number of amazing individuals during my time at the University of Minnesota. This thesis is the outcome of that wonderful association. First, I express my gratitude to my advisor, Professor. Shekhar, for being such a wonderful mentor. I am particularly grateful for his patient and steadfast support during anxious moments in Graduate School. I am indebted to him for helping me understand the principles behind the practice of science through several memorable experiences. I truly appreciate the con-fidence he instilled in me, the courage he has given me to face challenges in research and all the advise that he has been kind to give me through these years. He has inspired me with his balanced approach to any challenge, amazing wisdom and unique perspectives. I am blessed to have had him as my advisor and I consider having been his student a great honor. My sincere thanks to my dissertation committee members, Professor Kumar, Professor Srivastava, Professor Karypis, Professor Harvey and Professor Banerjee for their support, feedback and guidance. I am indebted to our collaborators from the U.S. Army Corps Engineers and the Na-tional Institute of Justice who gave me valuable insights into the societal applications of this research. I am particularly grateful for several valuable comments that shaped differ-ent parts of this thesis through publications in highly selective conferences and journals. I i
FOR THE DEGREE OF
, 2008
"... iAcknowledgements I wish to thank several people who, in some way, have helped me during the develop-ment of this thesis. First, I would like to express my gratitude to my supervisor, Prof. Shashi Shekhar for his continued encouragement and invaluable suggestions during this study. It has been a gre ..."
Abstract
- Add to MetaCart
(Show Context)
iAcknowledgements I wish to thank several people who, in some way, have helped me during the develop-ment of this thesis. First, I would like to express my gratitude to my supervisor, Prof. Shashi Shekhar for his continued encouragement and invaluable suggestions during this study. It has been a great pleasure to do research with him. I would like to thank members of my advisory committee Professors Jaideep Srivastava, Arindam Banerjee, and Sudipto Banerjee for their guidance and constructive and useful comments. This research also benefited from discussions with Prof. Daniel Boley. I sincerely would like to thank to US Army Corps of Engineers, Topographic Engi-neering Center Researchers James P. Rogers and Dr. James A. Shine with whom I have collaborated during my PhD research. Their conceptual and technical insights into my thesis work have been invaluable. I am very grateful to the all members of Spatial Databases and Data Mining group members at the Department of Computer Science for their discussions, critiques, and con-
Mining Co-locations Under Uncertainty
"... Abstract. A co-location pattern represents a subset of spatial features whose events tend to locate together in spatial proximity. The certain case of the co-location pattern has been investigated. However, location information of spatial features is often imprecise, aggregated, or error prone. Beca ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract. A co-location pattern represents a subset of spatial features whose events tend to locate together in spatial proximity. The certain case of the co-location pattern has been investigated. However, location information of spatial features is often imprecise, aggregated, or error prone. Because of the continuity nature of space, over-counting is a major problem. In the uncertain case, the problem becomes more challenging. In this paper, we propose a probabilistic participation index to measure co-location patterns based on the well-known possible world model. To avoid the exponential cost of calculating participation index from all possible worlds, we prove a lemma that allows for instance centric counting, avoids over-counting, and produces the same results as using possible world based counting. We use this property to develop efficient mining algorithms. We observed through both algebraic analysis and extensive experiments that the feature tree based algorithm outperforms uncertain Apriori algorithm by an order of magnitude not only for co-locations of large sizes but also for datasets with high level of uncertainty. This is an important insight in mining uncertainty co-locations. The uncertainty co-location mining can find which types of animals tend to locate in proximity. 1
