Search

K-nearest neighbors search (K-NNS) is a fundamental problem in many areas of machine learning and data mining. In an attempt to solve NNS problems by locality-sensitive hashing (LSH)-based algorithms and avoid the F1-trap, in this paper, we theoretically prove the necessity of maintaining close edges of each dimension, the independence of the partition location and the precision of retrieval. Based on the theories, a new method is proposed for constructing the hash function family. Then, a volume-balancing locality-sensitive hashing (VBLSH) algorithm is implemented. In this algorithm, the problem of constructing the hash function family is transformed into an optimization problem under a fixed number of hypercubes. By solving the optimization problem, different hyperplane intervals are set for each dimension to ensure that the hash functions can evenly segment the data space, where each segmented hypercube has a close volume and close side lengths. Finally, the analysis of the algorithm shows that the VBLSH algorithm has a high index coding efficiency and hash code calculation efficiency. The experimental results show that the VBLSH algorithm can simultaneously take into account the efficiency, precision and recall rate, and avoid the F1-trap; it shows excellent effectiveness on 4 datasets. [ABSTRACT FROM AUTHOR]

• Modified compact utility list (MCUL) storage structure is built. • Backward checking is performed to reduce the search space and runtime. • Forward checking is combined with building candidates to reduce the runtime. • Similar transactions or closed transactions are merged to reduce storage space. • Values of LA-prune and C-prune are updated as soon as an MCUL is pruned. Mining closed high utility itemsets (CHUIs) involves finding a representative set of HUIs that is usually smaller than that of HUIs but can generate the full HUIs without loss of information. Researchers have therefore shown interest in this problem, and many methods have been proposed for mining CHUIs effectively, of which CHUI-Miner and EFIM-Closed are the two most efficient algorithms. However, these face performance issues when mining CHUIs from sparse datasets. In this paper, we propose a method for the effective mining of CHUIs in both dense and sparse datasets. We first modify the compact utility list structure in the HMiner algorithm to reduce the mining time, and then develop backward and forward checking methods using the most recently explored CHUIs and combine this with candidate building for the next levels. Finally, we apply pruning strategies to reduce the search space for the generation of CHUIs. Our experimental results show that the proposed algorithm, called HMiner-Closed, is more efficient than the state-of-the-art algorithms for both dense and sparse datasets. [ABSTRACT FROM AUTHOR]

Mining high utility patterns is an important data mining problem that is formulated as finding patterns whose utilities are no less than a threshold. As the mining results are very sensitive to such a threshold, it is difficult for users to specify an appropriate one. An alternative formulation of the problem is to find the top- n high utility patterns. However, the second formulation is more challenging because the corresponding threshold is unknown in advance and the solution search space becomes even larger. When there are very long patterns prior algorithms simply cannot work to mine top- n high utility patterns even for very small n . This paper proposes a novel algorithm for mining top- n high utility patterns that are long. The proposed algorithm adopts an opportunistic pattern growth approach and proposes five opportunistic strategies for scalably maintaining shortlisted patterns, for efficiently computing utilities, and for estimating tight upper bounds to prune search space. Extensive experiments show that the proposed algorithm is 1 to 3 orders of magnitude more efficient than the state-of-the-art top- n high utility pattern mining algorithms, and it is even up to 2 orders of magnitude faster than high utility pattern mining algorithms that are tuned with an optimal threshold. [ABSTRACT FROM AUTHOR]

Data clustering aims to segment a database into groups of objects based on the similarity among these objects. Due to its unsupervised nature, the search for a good-quality solution can become a complex process. There is currently a wide range of clustering algorithms, and selecting the best one for a given problem can be a slow and costly process. In 1976, Rice formulated the Algorithm Selection Problem (ASP), which postulates that the algorithm performance can be predicted based on the structural characteristics of the problem. Meta-learning brings the concept of learning about learning; that is, the meta-knowledge obtained from the algorithm learning process allows the improvement of the algorithm performance. Meta-learning has a major intersection with data mining in classification problems, in which it is normally used to recommend algorithms. The present paper proposes new ways to obtain meta-knowledge for clustering tasks. Specifically, two contributions are explored here: (1) a new approach to characterize clustering problems based on the similarity among objects; and (2) new methods to combine internal indices for ranking algorithms based on their performance on the problems. Experiments were conducted to evaluate the recommendation quality. The results show that the new meta-knowledge provides high-quality algorithm selection for clustering tasks. [ABSTRACT FROM AUTHOR]