Current Search:  similarity search (x)

View All Items

  • CSV Spreadsheet
(1 - 2 of 2)
LEARNING TECHNIQUES FOR INFORMATION RETRIEVAL AND MINING IN HIGH-DIMENSIONAL DATABASES
Title
LEARNING TECHNIQUES FOR INFORMATION RETRIEVAL AND MINING IN HIGH-DIMENSIONAL DATABASES.
Creator
Cheng, Hao, Hua, Kien A., University of Central Florida
Abstract / Description
The main focus of my research is to design effective learning techniques for information retrieval and mining in high-dimensional databases. There are two main aspects in the retrieval and mining research: accuracy and efficiency. The accuracy problem is how to return results which can better match the ground truth, and the efficiency problem is how to evaluate users' requests and execute learning algorithms as fast as possible. However, these problems are non-trivial because of the...
Show moreThe main focus of my research is to design effective learning techniques for information retrieval and mining in high-dimensional databases. There are two main aspects in the retrieval and mining research: accuracy and efficiency. The accuracy problem is how to return results which can better match the ground truth, and the efficiency problem is how to evaluate users' requests and execute learning algorithms as fast as possible. However, these problems are non-trivial because of the complexity of the high-level semantic concepts, the heterogeneous natures of the feature space, the high dimensionality of data representations and the size of the databases. My dissertation is dedicated to addressing these issues. Specifically, my work has five main contributions as follows. The first contribution is a novel manifold learning algorithm, Local and Global Structures Preserving Projection (LGSPP), which defines salient low-dimensional representations for the high-dimensional data. A small number of projection directions are sought in order to properly preserve the local and global structures for the original data. Specifically, two groups of points are extracted for each individual point in the dataset: the first group contains the nearest neighbors of the point, and the other set are a few sampled points far away from the point. These two point sets respectively characterize the local and global structures with regard to the data point. The objective of the embedding is to minimize the distances of the points in each local neighborhood and also to disperse the points far away from their respective remote points in the original space. In this way, the relationships between the data in the original space are well preserved with little distortions. The second contribution is a new constrained clustering algorithm. Conventionally, clustering is an unsupervised learning problem, which systematically partitions a dataset into a small set of clusters such that data in each cluster appear similar to each other compared with those in other clusters. In the proposal, the partial human knowledge is exploited to find better clustering results. Two kinds of constraints are integrated into the clustering algorithm. One is the must-link constraint, indicating that the involved two points belong to the same cluster. On the other hand, the cannot-link constraint denotes that two points are not within the same cluster. Given the input constraints, data points are arranged into small groups and a graph is constructed to preserve the semantic relations between these groups. The assignment procedure makes a best effort to assign each group to a feasible cluster without violating the constraints. The theoretical analysis reveals that the probability of data points being assigned to the true clusters is much higher by the new proposal, compared to conventional methods. In general, the new scheme can produce clusters which can better match the ground truth and respect the semantic relations between points inferred from the constraints. The third contribution is a unified framework for partition-based dimension reduction techniques, which allows efficient similarity retrieval in the high-dimensional data space. Recent similarity search techniques, such as Piecewise Aggregate Approximation (PAA), Segmented Means (SMEAN) and Mean-Standard deviation (MS), prove to be very effective in reducing data dimensionality by partitioning dimensions into subsets and extracting aggregate values from each dimension subset. These partition-based techniques have many advantages including very efficient multi-phased pruning while being simple to implement. They, however, are not adaptive to different characteristics of data in diverse applications. In this study, a unified framework for these partition-based techniques is proposed and the issue of dimension partitions is examined in this framework. An investigation of the relationships of query selectivity and the dimension partition schemes discovers indicators which can predict the performance of a partitioning setting. Accordingly, a greedy algorithm is designed to effectively determine a good partitioning of data dimensions so that the performance of the reduction technique is robust with regard to different datasets. The fourth contribution is an effective similarity search technique in the database of point sets. In the conventional model, an object corresponds to a single vector. In the proposed study, an object is represented by a set of points. In general, this new representation can be used in many real-world applications and carries much more local information, but the retrieval and learning problems become very challenging. The Hausdorff distance is the common distance function to measure the similarity between two point sets, however, this metric is sensitive to outliers in the data. To address this issue, a novel similarity function is defined to better capture the proximity of two objects, in which a one-to-one mapping is established between vectors of the two objects. The optimal mapping minimizes the sum of distances between each paired points. The overall distance of the optimal matching is robust and has high retrieval accuracy. The computation of the new distance function is formulated into the classical assignment problem. The lower-bounding techniques and early-stop mechanism are also proposed to significantly accelerate the expensive similarity search process. The classification problem over the point-set data is called Multiple Instance Learning (MIL) in the machine learning community in which a vector is an instance and an object is a bag of instances. The fifth contribution is to convert the MIL problem into a standard supervised learning in the conventional vector space. Specially, feature vectors of bags are grouped into clusters. Each object is then denoted as a bag of cluster labels, and common patterns of each category are discovered, each of which is further reconstructed into a bag of features. Accordingly, a bag is effectively mapped into a feature space defined by the distances from this bag to all the derived patterns. The standard supervised learning algorithms can be applied to classify objects into pre-defined categories. The results demonstrate that the proposal has better classification accuracy compared to other state-of-the-art techniques. In the future, I will continue to explore my research in large-scale data analysis algorithms, applications and system developments. Especially, I am interested in applications to analyze the massive volume of online data.
Show less
Date Issued
2009
Identifier
CFE0002882, ucf:48022
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0002882
Hashing for Multimedia Similarity Modeling and Large-Scale Retrieval
Title
Hashing for Multimedia Similarity Modeling and Large-Scale Retrieval.
Creator
Li, Kai, Hua, Kien, Qi, GuoJun, Hu, Haiyan, Wang, Chung-Ching, University of Central Florida
Abstract / Description
In recent years, the amount of multimedia data such as images, texts, and videos have been growing rapidly on the Internet. Motivated by such trends, this thesis is dedicated to exploiting hashing-based solutions to reveal multimedia data correlations and support intra-media and inter-media similarity search among huge volumes of multimedia data.We start by investigating a hashing-based solution for audio-visual similarity modeling and apply it to the audio-visual sound source localization...
Show moreIn recent years, the amount of multimedia data such as images, texts, and videos have been growing rapidly on the Internet. Motivated by such trends, this thesis is dedicated to exploiting hashing-based solutions to reveal multimedia data correlations and support intra-media and inter-media similarity search among huge volumes of multimedia data.We start by investigating a hashing-based solution for audio-visual similarity modeling and apply it to the audio-visual sound source localization problem. We show that synchronized signals in audio and visual modalities demonstrate similar temporal changing patterns in certain feature spaces. We propose to use a permutation-based random hashing technique to capture the temporal order dynamics of audio and visual features by hashing them along the temporal axis into a common Hamming space. In this way, the audio-visual correlation problem is transformed into a similarity search problem in the Hamming space. Our hashing-based audio-visual similarity modeling has shown superior performances in the localization and segmentation of sounding objects in videos.The success of the permutation-based hashing method motivates us to generalize and formally define the supervised ranking-based hashing problem, and study its application to large-scale image retrieval. Specifically, we propose an effective supervised learning procedure to learn optimized ranking-based hash functions that can be used for large-scale similarity search. Compared with the randomized version, the optimized ranking-based hash codes are much more compact and discriminative. Moreover, it can be easily extended to kernel space to discover more complex ranking structures that cannot be revealed in linear subspaces. Experiments on large image datasets demonstrate the effectiveness of the proposed method for image retrieval.We further studied the ranking-based hashing method for the cross-media similarity search problem. Specifically, we propose two optimization methods to jointly learn two groups of linear subspaces, one for each media type, so that features' ranking orders in different linear subspaces maximally preserve the cross-media similarities. Additionally, we develop this ranking-based hashing method in the cross-media context into a flexible hashing framework with a more general solution. We have demonstrated through extensive experiments on several real-world datasets that the proposed cross-media hashing method can achieve superior cross-media retrieval performances against several state-of-the-art algorithms.Lastly, to make better use of the supervisory label information, as well as to further improve the efficiency and accuracy of supervised hashing, we propose a novel multimedia discrete hashing framework that optimizes an instance-wise loss objective, as compared to the pairwise losses, using an efficient discrete optimization method. In addition, the proposed method decouples the binary codes learning and hash function learning into two separate stages, thus making the proposed method equally applicable for both single-media and cross-media search. Extensive experiments on both single-media and cross-media retrieval tasks demonstrate the effectiveness of the proposed method.
Show less
Date Issued
2017
Identifier
CFE0006759, ucf:51840
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0006759