Search

In this paper, we address the challenge of creating candidate sets for large-scale Traveling Salesman Problem (TSP) instances, where choosing a subset of edges is crucial for efficiency. Traditional methods for improving tours, such as local searches and heuristics, depend greatly on the quality of these candidate sets but often struggle in large-scale situations due to insufficient edge coverage or high time complexity. We present a new heuristic based on fuzzy clustering, designed to produce high-quality candidate sets with nearly linear time complexity. Thoroughly tested on benchmark instances, including VLSI and Euclidean types with up to 316,000 nodes, our method consistently outperforms traditional and current leading techniques for large TSPs. Our heuristic's tours encompass nearly all edges of optimal or best-known solutions, and its candidate sets are significantly smaller than those produced with the POPMUSIC heuristic. This results in faster execution of subsequent improvement methods, such as Helsgaun's Lin–Kernighan heuristic and evolutionary algorithms. This substantial enhancement in computation time and solution quality establishes our method as a promising approach for effectively solving large-scale TSP instances. [ABSTRACT FROM AUTHOR]

Network anomaly detection plays a vital role in safeguarding network security. However, the existing network anomaly detection task is typically based on the one-class zero-positive scenario. This approach is susceptible to overfitting during the training process due to discrepancies in data distribution between the training set and the test set. This phenomenon is known as prediction drift. Additionally, the rarity of anomaly data, often masked by normal data, further complicates network anomaly detection. To address these challenges, we propose the PUNet network, which ingeniously combines the strengths of traditional machine learning and deep learning techniques for anomaly detection. Specifically, PUNet employs a reconstruction-based autoencoder to pre-train normal data, enabling the network to capture potential features and correlations within the data. Subsequently, PUNet integrates a sampling algorithm to construct a pseudo-label candidate set among the outliers based on the reconstruction loss of the samples. This approach effectively mitigates the prediction drift problem by incorporating abnormal samples. Furthermore, PUNet utilizes the CatBoost classifier for anomaly detection to tackle potential data imbalance issues within the candidate set. Extensive experimental evaluations demonstrate that PUNet effectively resolves the prediction drift and data imbalance problems, significantly outperforming competing methods. [ABSTRACT FROM AUTHOR]

In this paper, we address the challenge of creating candidate sets for large-scale Traveling Salesman Problem (TSP) instances, where choosing a subset of edges is crucial for efficiency. Traditional methods for improving tours, such as local searches and heuristics, depend greatly on the quality of these candidate sets but often struggle in large-scale situations due to insufficient edge coverage or high time complexity. We present a new heuristic based on fuzzy clustering, designed to produce high-quality candidate sets with nearly linear time complexity. Thoroughly tested on benchmark instances, including VLSI and Euclidean types with up to 316,000 nodes, our method consistently outperforms traditional and current leading techniques for large TSPs. Our heuristic’s tours encompass nearly all edges of optimal or best-known solutions, and its candidate sets are significantly smaller than those produced with the POPMUSIC heuristic. This results in faster execution of subsequent improvement methods, such as Helsgaun’s Lin–Kernighan heuristic and evolutionary algorithms. This substantial enhancement in computation time and solution quality establishes our method as a promising approach for effectively solving large-scale TSP instances.

The deep neural network is similar to the biological neural network, so it has the ability of high efficiency and accurate extraction of the deep hidden features of information, can learn multiple layers of abstract features, and can learn more about cross-domain, multi-source and heterogeneous content information. This paper presents an extraction feature based on multi-user-project combined deep neural network, self-learning and other advantages to achieve the model of personalized information. This model does deep neural network self-learning and extraction based on the input multi-source heterogeneous data characteristics，fuses collaborative filtering wide personalization to generate candidate sets, and then through two times of model self-learning produces a sort set. Finally，it can achieve accurate, real-time, and personalized recommendations. The experimental results show that the model can self-learn and extract the user′s implicit feature well, and it can solve the problems of sparse and new items of traditional recommendation system to some extent, and realize more accurate, real-time and personalized recommendation.

Deep neural networks (DNNs) have promoted much of the recent progress in hyperspectral image (HSI) classification, which depends on extensive labeled samples and deep network structure and has achieved surprisingly good generalization capacity. However, due to the expensive labeling cost, the labeled samples are scarce in most practice cases, which causes these DNN-based methods to be prone to over-fitting and influences the classification result. To mitigate this problem, we present a clustering-inspired active learning method for enhancing the HSI classification result, which mainly contributes to two aspects. On one hand, the modified clustering by fast search and find of peaks clustering method is utilized to select highly informative and diverse samples from unlabeled samples in the candidate set for manual labeling, which empowers us to appropriately augment the limited training set (i.e., labeled samples) and thus improves the generalization capacity of the baseline DNN model. On the other hand, another K-means clustering-based pseudo-labeling scheme is utilized to pre-train the DNN model with all samples in the candidate set. By doing this, the pre-trained model can be effectively generalized to unlabeled samples in the testing set after being fine tuned-based on the augmented training set. The experiment accuracies on two benchmark HSI datasets show the effectiveness of the proposed method.

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In order to further improve the speed of the association rules mining frequent sets and optimize the execution performance of the algorithm, this paper proposed an association rule mining algorithm based on improved memory structure. Based on the Spark distributed framework, the proposed algorithm mined frequent sets in parallel. It used the Bloom filter to store the project in the mining process, and simplified the operation of the transaction set and the candidate set, so as to optimize the speed of mining frequent sets and save the computing resources. Compared with the YAFIM and the MR-Apriori algorithm, the proposed algorithm has a significant improvement in the efficiency of mining frequent sets under the condition of occupying less memory. The algorithm can not only improve the mining speed and reduce the memory pressure, but also has good scalability, so that the algorithm can be applied to larger data sets and clusters to optimize the performance. [ABSTRACT FROM AUTHOR]

We propose a contraction transformation algorithm to plan a complete coverage trajectory for a mobile robot to accomplish specific types of missions based on the Arnold dynamical system. First, we construct a chaotic mobile robot by combining the variable z of the Arnold equation and the kinematic equation of the robot. Second, we construct the candidate sets including the initial points with a relatively high coverage rate of the constructed mobile robot. Then the trajectory is contracted to the current position of the robot based on the designed contraction transformation strategy, to form a continuous complete coverage trajectory to execute the specific types of missions. Compared with the traditional method, the designed algorithm requires no obstacle avoidance to the boundary of the given workplace, possesses a high coverage rate, and keeps the chaotic characteristics of the produced coverage trajectory relatively unchanged, which enables the robot to accomplish special missions with features of completeness, randomness, or unpredictability. [ABSTRACT FROM AUTHOR]

Research on wireless sensor network (WSN) has mainly involved the use of a portable and limited power source, namely batteries, to power the sensors. Without energy, a sensor is essentially useless and can not contribute to the utility of the network as a whole. As a result, considerable research effort has been devoted to designing energy efficient networking protocols to maximize network life. For this reason, a new routing concept called opportunistic routing has emerged whose objective is to address the resource management problem in order to maximize the lifetime of the network. In other words, opportunistic routing is a new theme in the WSNs, as it is an approach to take into account the environment of a node as it is and not as perceived by the network. Our main objective in this work will be to present a new idea to calculate the list of candidates to minimize energy consumption and increased the life time of network. And we proposed a new opportunistic routing protocols called energy-efficient opportunistic routing using a new forward list (EEOR-FL) aiming to balance energy consumption and maximize the network lifetime. EEOR-FL uses a new method in order to to calculate the list of candidates. The obtained results indicate that EEOR-FL can effectively balance the energy consumption and therefore can lengthen the network lifetime. [ABSTRACT FROM AUTHOR]

Contour detection has been extensively investigated as a fundamental problem in computer vision. In this study, a biologically-inspired candidate weighting framework is proposed for the challenging task of detecting meaningful contours. In contrast to previous models that detect contours from pixels, a modified superpixel generation processing is proposed to generate a contour candidate set and then weigh the candidates by extracting hierarchical visual cues. We extract the low-level visual local cues to weigh the contour intrinsic property and mid-level visual cues on the basis of Gestalt principles for weighting the contour grouping constraint. Experimental results tested on the BSDS benchmark show that the proposed framework exhibits promising performances to capture meaningful contours in complex scenes.

At present, most of the researches on geometric change detection of vector data, they store the change detection results in the database, so they pay more attention to the accuracy of results, but not to the speed of processing. Nowadays, many applications require real-time change detection on vector data and rapid presentation of the result. Although the existing algorithms use spatial index technology to improve the processing speed, the processing time is still beyond the range that people can bear. In order to reduce processing time, this paper takes the vector surface feature set as the research object, trying to reduce the redundancy of the candidate set that seriously affects the efficiency of change detection. Based on the regular use of spatial index created with geometric Minimum Bounding Rectangle, this paper uses geometric shrinkage technique and precise query technique to reduce the size of the candidate set for detection, so as to achieve the goal of speeding up. Finally, using five years of farmland data and resident data from Ezhou City, Hubei Province, China, a change detection experiment was conducted. The experiment proved that the geometric shrinkage and precise query techniques can effectively improve the processing speed. [ABSTRACT FROM AUTHOR]

Opportunistic routing is a promising research area in the context of wireless network communications. Security and trustworthy of routing in this field, however, needs to be considerably researched . In this paper, a novel trust establishment algorithm is proposed, designed, and implemented specifically for opportunistic routing protocols which benefits from direct interactions between wireless nodes. The proposed trust model benefits from a novel watchdog mechanism considering not only forwarding behaviour of nodes but also the quality of links between them. Furthermore, three different metrics for next hop selection is introduced enabling nodes to select their next hop forwarders more sophisticatedly using quality of links, geographical location of nodes, and their trust level. Extensive simulation results represent that proposed model can significantly improve the performance of network communications when malicious nodes try to collapse the system. [ABSTRACT FROM AUTHOR]

Test data generation is a key issue in the field of software testing. Adaptive random testing (ART) method has been proposed by Chen et al. to improve the fault-revealing ability of random testing. In the paper, we are mainly concerned with the partitioning-based adaptive random testing and present a new ART based on two-point partitioning. In the new algorithm, the current max-area region is partitioned by the midpoint of two points instead of a single point. The first point is randomly generated, and the second point is picked out from the candidate set according to the farthest distance criterion. In order to compare our algorithm with other two well-known algorithms, the experiments for the case of two-dimension are performed. The results show that our ART-TPP algorithm has a positive improvement for the other two, i.e. ART-RP and ART-BP. Moreover, the appropriate size of candidate set is determined as 2 or 3 based on our sensitivity analysis. [ABSTRACT FROM AUTHOR]

To overcome the default of single search tendency, the ants in the colony are divided into several sub-groups. The ants in different subgroups have different trail information and expectation coefficients. The simulated annealing method is introduced to the algorithm. Through setting the temperature changing with the iterations, after each turn of tours, the solution set obtained by the ants is taken as the candidate set. The update set is obtained by adding the solutions in the candidate set to the previous update set with the probability determined by the temperature. The solutions in the candidate set are used to update the trail information. In each turn of updating, the current best solution is also used to enhance the trail information on the current best route. The trail information is reset when the algorithm is in stagnation state. The computer experiments demonstrate that the proposed algorithm has higher stability and convergence speed. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) have promoted much of the recent progress in hyperspectral image (HSI) classification, which depends on extensive labeled samples and deep network structure and has achieved surprisingly good generalization capacity. However, due to the expensive labeling cost, the labeled samples are scarce in most practice cases, which causes these DNN-based methods to be prone to over-fitting and influences the classification result. To mitigate this problem, we present a clustering-inspired active learning method for enhancing the HSI classification result, which mainly contributes to two aspects. On one hand, the modified clustering by fast search and find of peaks clustering method is utilized to select highly informative and diverse samples from unlabeled samples in the candidate set for manual labeling, which empowers us to appropriately augment the limited training set (i.e., labeled samples) and thus improves the generalization capacity of the baseline DNN model. On the other hand, another K-means clustering-based pseudo-labeling scheme is utilized to pre-train the DNN model with all samples in the candidate set. By doing this, the pre-trained model can be effectively generalized to unlabeled samples in the testing set after being fine tuned-based on the augmented training set. The experiment accuracies on two benchmark HSI datasets show the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Contour detection has been extensively investigated as a fundamental problem in computer vision. In this study, a biologically-inspired candidate weighting framework is proposed for the challenging task of detecting meaningful contours. In contrast to previous models that detect contours from pixels, a modified superpixel generation processing is proposed to generate a contour candidate set and then weigh the candidates by extracting hierarchical visual cues. We extract the low-level visual local cues to weigh the contour intrinsic property and mid-level visual cues on the basis of Gestalt principles for weighting the contour grouping constraint. Experimental results tested on the BSDS benchmark show that the proposed framework exhibits promising performances to capture meaningful contours in complex scenes.