The goal of the high-utility itemset mining task is to discover combinations of items which that yield high profits from transactional databases. HUIM is a useful tool for retail stores to analyze customer behaviors. ...
详细信息
ISBN:
(纸本)9783030732790;9783030732806
The goal of the high-utility itemset mining task is to discover combinations of items which that yield high profits from transactional databases. HUIM is a useful tool for retail stores to analyze customer behaviors. However, it ignores the categorization of items. To solve this issue, the ML-HUI Miner algorithm was presented. It combines item taxonomy with the HUIM task and is able to discover insightful itemsets, which are not found in traditionalHUIMapproaches. Although ML-HUI Miner is efficient in discovering itemsets from multiple abstraction levels, it is a sequential algorithm. Thus, it cannot utilize the powerful multi-core processors, which are currently available widely. This paper addresses this issue by extending the algorithm into a multi-core version, called the MCML-Miner algorithm (multi-coremulti-Level high-utility itemset Miner), to help reduce significantly the mining time. Each level in the taxonomy will be assigned a separate processor core to explore concurrently. Experiments on real-world datasets show that theMCML-Miner up to several folds faster than the original algorithm.
Among the useful tools for the retail stores to analyze their customer behaviors is through the task of mining high-utility itemset (HUIM), which is to reveal the combinations of items which offer high. However, most ...
详细信息
ISBN:
(纸本)9783030880811;9783030880804
Among the useful tools for the retail stores to analyze their customer behaviors is through the task of mining high-utility itemset (HUIM), which is to reveal the combinations of items which offer high. However, most of them different abstraction levels of items. The CLH-Miner algorithm was presented to solve this problem. It adopts categorization of items with the HUIM to discover interesting itemsets not contained in traditional HUIM approaches. Whereas CLH-Miner discovers itemsets from different levels of abstraction efficiently, the algorithm is sequential. It cannot, therefore, use powerful, easily available, multi-core processors. This work tackles this drawback through the use of a parallel method called the pCLH-Miner algorithm to significantly reduce mining times. The algorithm proposes a way to split the search space into separate parts and assign them to each different core. The pCLH-miner is shown to high efficiency compared CLH-Miner by experiments on real-world databases.
This paper addresses the problem of big association rule mining using an evolutionary approach. The mimetic method has been successfully applied to small and medium size databases. However, when applied on larger data...
详细信息
ISBN:
(纸本)9789811064876;9789811064869
This paper addresses the problem of big association rule mining using an evolutionary approach. The mimetic method has been successfully applied to small and medium size databases. However, when applied on larger databases, the performance of this method becomes an important issue and current algorithms have very long execution times. Modern CPU/GPU architectures are composed of many cores, which are massively threaded and provide a large amount of computing power, suitable for improving the performance of optimization techniques. The parallelization of such method on GPU architecture is thus promising to deal with very large datasets in real time. In this paper, an approach is proposed where the rule evaluation process is parallelized on GPU, while the generation of rules is performed on a multi-core CPU. Furthermore, an intelligent strategy is proposed to partition the search space of rules in several independent sub-spaces to allow multiple CPU cores to explore the search space efficiently and without performing redundant work. Experimental results reveal that the suggested approach outperforms the sequential version by up to at 600 times for large datasets. Moreover, it outperforms the-state-of-the-art high performance computing based approaches when dealing with the big WebDocs dataset.
Massive individuals identification is a challenging problem in modern society. Particularly, finger-vein recognition is an emerging biometric technique with several advantages, especially in terms of security against ...
详细信息
ISBN:
(纸本)9781728156132
Massive individuals identification is a challenging problem in modern society. Particularly, finger-vein recognition is an emerging biometric technique with several advantages, especially in terms of security against forgery. In this paper, we propose a hybrid parallel matching process for finger-vein recognition under two different parallel platforms, by using the Local Line Binary Pattern descriptor and Hamming distance. Our proposal aims to reduce the computation time of the matching process for massive individuals identification by using finger-vein patterns. Extensive evaluation shows that our approach obtains a high speed-up under a multi-core platform and a close to linear behavior for a multi-node platform. The results with our hybrid parallel system show that it is suitable for real-time individuals identification, achieving a speed-up up to 129.98x. To the best of our knowledge, our work is the first implementation of finger-vein recognition under a parallel platform, which is the main contribution of this paper.
We describe a parallel algorithm for solving parity games, with applications in, e.g., modal mu-calculus model checking with arbitrary alternations, and (branching) bisimulation checking. The algorithm is based on Jur...
详细信息
We describe a parallel algorithm for solving parity games, with applications in, e.g., modal mu-calculus model checking with arbitrary alternations, and (branching) bisimulation checking. The algorithm is based on Jurdzinski's Small Progress Measures. Actually, this is a class of algorithms, depending on a selection heuristics. Our algorithm operates lock-free, and mostly wait-free (except for infrequent termination detection), and thus allows maximum parallelism. Additionally, we conserve memory by avoiding storage of predecessor edges for the parity graph through strictly forward-looking heuristics. We evaluate our multi-core implementation's behaviour on parity games obtained from mu-calculus model checking problems for a set of communication protocols, randomly generated problem instances, and parametric problem instances from the literature.
暂无评论