Random Swap Research Articles

Distributed random swap: An efficient algorithm for minimum sum-of-squares clustering

The clustering model known as Minimum Sum-of-Squares Clustering (MSSC) is widely used, with the popular k-means algorithm serving as its local minimizer. It is well-known that solutions of k-means can result in substantial deviations from the true global optimum of MSSC. While numerous heuristics and metaheuristics have been proposed to overcome this limitation, none have gained dominant acceptance in academic literature. This is likely related to challenges such as intricate implementations and a multitude of tunable parameters.In this paper, we dispute the belief that simplifying an algorithm for MSSC inherently means sacrificing quality. We present the Distributed Random Swap (DRS-means) algorithm, which is designed to enhance clustering performance for the MSSC problem. This algorithm can be interpreted as an iterative method that refines the solution generated by the k-means algorithm during each iteration. The enhancement is achieved by selecting points based on specific probability distributions. These distributions are carefully designed to improve and speed up the exploration phase. The proposed algorithm is straightforward to implement. DRS-means offers a user-friendly solution with state-of-the-art results, making it suitable for a wide range of research fields.

An Empirical Study on the Application of Data Augmentation Techniques to Enhance the Performance of CNN Models in Spam Detection

Spam detection is a critical task in cybersecurity, aiming to filter out unsolicited and potentially harmful communications. This study investigates the impact of various data augmentation techniques on enhancing the performance of Convolutional Neural Network (CNN) models for spam detection. Utilizing the Enron Email Dataset, we implemented several augmentation methods, including synonym replacement, random insertion, random swap, random deletion, back translation, and noise addition. Our results indicate significant performance improvements with these techniques. The baseline CNN model achieved an accuracy of 87.5%, precision of 85.2%, recall of 83.7%, and F1-score of 84.4%. The application of back translation, the most effective technique, increased accuracy to 90.3% and F1-score to 88.0%. These findings demonstrate the potential of data augmentation in improving spam detection systems, providing a robust foundation for future research. The study also highlights the importance of combining augmentation techniques and adapting them to different languages and real-world scenarios for even greater performance gains.

Stochastic inversion of fracture networks using the reversible jump Markov chain Monte Carlo algorithm

Characterization of fracture networks is essential in the production optimization or storage calculation for enhanced geothermal systems or geologic carbon storage. A novel inversion approach is proposed for estimating the fracture networks in this research. The discrete fracture network technique is adopted to probabilistically describe various fracture parameters such as trace length, midpoint position or azimuthal angle. The reversible jump Markov chain Monte Carlo algorithm is applied to explore the target posterior distribution of model parameters of differing dimensionality, in which the number of fractures is assumed to be unknown. More specifically, the birth-death strategy is utilized to perturb the fracture number iteratively in the sampling process. The proposed methodology is applied with two different types of observational datasets, namely the head records from steady-state flow simulation and the acoustic impedance obtained from seismic inversion. The sampling results can successfully recover the fracture geometry in the observed domain, and the number of fractures in the system can be retrieved as well. Benchmarked on multiple Markov chain trials, the technique of parallel tempering can greatly improve the convergence efficiency and increase the diversity of sampled posterior models, through the random swapping of model states across the whole temperature ladder.

Towards a robust scale‐free network in internet of health things against multiple attacks using an inter‐core based reconnection strategy

SummaryWireless sensor networks (WSNs) have attained a great attraction of researchers in the recent years. In these networks, many structures are considered that have different properties. This article offers a unique approach, the inter‐core based reconnection strategy (ICRS), which is intended to improve the robustness of Scale‐Free Networks (SFNs) in the setting of wireless sensor networks (WSNs), with a special emphasis on the Internet of Health Things (IoHT) network. SFNs' vulnerabilities to malicious assaults while remaining resilient to random attacks. The proposed ICRS overcomes this issue by offering a novel reconnection approach that employs separate edges between network centers. Destructive assaults that have a significant impact on network connectivity, emphasizing the importance of a robust network that can resist a variety of attacks. ICRS is positioned as a solution that optimizes the network via reconnection techniques, changing it into an onion‐like structure with increased robustness. The simulation results depict that ICRS outperforms the existing algorithms in terms of robustness enhancement. The results show that ICRS performs 48%, 29%, 22%, and 16% better than Barabasi Albert (BA), Hill Climbing (HC), Simulated Annealing (SA), Random Edge Swap Mechanism (RESM), and Robustness Strategy (ROSE), respectively.

Exploring the potential of data augmentation in poetry generation with small-scale corpora

Poetry generation is a complex task in the field of natural language processing, especially when working with small datasets. Data augmentation techniques have been shown to be an effective way to improve the performance of deep learning models in various tasks, including image classification and speech recognition. Therefore, this study focuses on exploring the impact of four different data augmentation methods - Synonym Replacement, Random Insertion, Random Swap, and Random Deletion - on the performance of poetry generation with a small poetry dataset. The results of the study reveal that Random Insertion performed well in terms of Bilingual Evaluation Understudy (BLEU), Recall-Oriented Understudy for Gisting Evaluation (ROUGE), and manual evaluation when compared to other data augmentation techniques. Synonym Replacement performed poorly in all three evaluations. This study confirms the potential value of data augmentation technology in poetry generation tasks and provides innovative perspectives and directions for future research in this area. Data augmentation can be employed to help address the problem of limited data in poetry generation tasks and enhance the efficiency of deep learning models. Future research could focus on exploring more advanced data augmentation techniques and their impact on poetry generation tasks.

Color image ROI encryption algorithm based on a novel 4D hyperchaotic system

The significance of safeguarding the security of image information has escalated significantly, owing to the exponential proliferation of digital images containing sensitive information being disseminated on the Internet. In this paper, we first propose a novel 4D hyperchaotic system and design a new image encryption algorithm in conjunction with the hyperchaotic system. The algorithm uses a split random swap permutation method to permute the image and combines the S-box to diffuse the image. To improve the diffusivity of this encryption algorithm, a cross-random diffusion method is designed to diffuse the image again. Then, we propose a region of interest (ROI) encryption scheme for images. This scheme can automatically identify irregular privacy targets in images and encrypt them. To ensure the security of the region of interest location information during transmission, the scheme compresses the location information of the privacy target using a run-length encoding technique and then embeds the compressed data into the ciphertext image using reversible steganography based on histogram shift. The experimental results and security analysis unequivocally demonstrate that the image encryption algorithm proposed in this paper exhibits robust resistance against a wide array of attacks, thereby ensuring a high level of security. Additionally, the devised image ROI encryption scheme effectively safeguards diverse privacy targets.

Improving Clustering Accuracy of K-Means and Random Swap by an Evolutionary Technique Based on Careful Seeding

K-Means is a “de facto” standard clustering algorithm due to its simplicity and efficiency. K-Means, though, strongly depends on the initialization of the centroids (seeding method) and often gets stuck in a local sub-optimal solution. K-Means, in fact, mainly acts as a local refiner of the centroids, and it is unable to move centroids all over the data space. Random Swap was defined to go beyond K-Means, and its modus operandi integrates K-Means in a global strategy of centroids management, which can often generate a clustering solution close to the global optimum. This paper proposes an approach which extends both K-Means and Random Swap and improves the clustering accuracy through an evolutionary technique and careful seeding. Two new algorithms are proposed: the Population-Based K-Means (PB-KM) and the Population-Based Random Swap (PB-RS). Both algorithms consist of two steps: first, a population of J candidate solutions is built, and then the candidate centroids are repeatedly recombined toward a final accurate solution. The paper motivates the design of PB-KM and PB-RS, outlines their current implementation in Java based on parallel streams, and demonstrates the achievable clustering accuracy using both synthetic and real-world datasets.

Bug report priority prediction using social and technical features

SummarySoftware stakeholders report bugs in issue tracking system (ITS) with manually labeled priorities. However, the lack of knowledge and standard for prioritization may cause stakeholders to mislabel the priorities. In response, priority predictors are actively developed to support them. Prior studies trained machine learners based on textual similarity, categorical, and numeric technical features of bug reports. Most models were validated by time‐insensitive approaches, and they were producing suboptimal results for practical usage. While they ignored the social aspects of ITS, the technical aspects were also limited in surface features of bug reports. To better model the bug report, we extract their topic and most similar code structures. Since ITS bridges users and developers as the main contributors, we also integrate their experience, sentiment, and socio‐technical features to construct a new dataset. Then, we perform two‐classed and multiclassed bug priority prediction based on the dataset. We also introduce adversarial training using generated training data with random word swap and random word deletion. We validate our model in within‐project, cross‐project, and time‐wise scenarios, and it outperforms the two baselines by up to 15% in area under curve‐receiver operating characteristics (AUC‐ROC) and 19% in Matthews correlation coefficient (MCC). We reveal involving contributor (i.e., assignee and reporter) features such as sentiment that could boost prediction performance. Finally, we test statistically the mean and distribution of the features that reflect the differences in social and technical aspects (e.g., quality of communication and resource distribution) between high and low priority reports. In conclusion, we suggest that researchers should consider both social and technical aspects of ITS in bug report priority prediction and introduce adversarial training to boost model performance.

An adaptive differential evolution algorithm with DBSCAN for the integrated slab allocation problem in steel industry

In this study, an integrated slab allocation problem (ISAP) is investigated, aiming to optimise inventory management in the steel production process of hot rolling. Different from previous studies, three sets of decisions are made simultaneously. The first is to allocate open-order slabs to orders to reduce inventory slabs and improve slab utilisation. The second is to reallocate customer-order slabs to orders to shorten the order completion time and improve customer satisfaction. The third is to group the remaining open-order slabs into virtual orders to ensure continuous production and reduce inventory pressure. The problem is formulated as an original integer programming mathematical model. A novel improved differential evolution algorithm with self-adaptive mutation and parameter selection strategies is proposed to solve the ISAP. In particular, a special encoding and decoding method is designed to address the issue of transformation between the slab allocation scheme and algorithm individual. A density-based spatial clustering of applications with noise algorithm and random swap local search are embedded into the framework of the differential evolution algorithm. Finally, experiments are conducted to evaluate the proposed algorithm, and the numerical results demonstrate its superiority in solving the ISAP.

Two Medoid-Based Algorithms for Clustering Sets

This paper proposes two algorithms for clustering data, which are variable-sized sets of elementary items. An example of such data occurs in the analysis of a medical diagnosis, where the goal is to detect human subjects who share common diseases to possibly predict future illnesses from previous medical history. The first proposed algorithm is based on K-medoids and the second algorithm extends the random swap algorithm, which has proven to be capable of efficient and careful clustering; both algorithms depend on a distance function among data objects (sets), which can use application-sensitive weights or priorities. The proposed distance function makes it possible to exploit several seeding methods that can improve clustering accuracy. A key factor in the two algorithms is their parallel implementation in Java, based on functional programming using streams and lambda expressions. The use of parallelism smooths out the O(N2) computational cost behind K-medoids and clustering indexes such as the Silhouette index and allows for the handling of non-trivial datasets. This paper applies the algorithms to several benchmark case studies of sets and demonstrates how accurate and time-efficient clustering solutions can be achieved.

Improving the performance of automatic short answer grading using transfer learning and augmentation

The task of grading answers ranging from one phrase to one paragraph using computational techniques is known as Automated Short Answer Grading (ASAG). The performance of existing systems is not good enough due to limited data and the lack of availability of data in many domains. Many ASAG systems were developed as an outcome of the active research in this field. This study builds an effective system for grading short answers in the programming domain by leveraging Pre-trained Language Models and Text Augmentation. We fine-tuned three-sentence transformer models on the SPRAG corpus with five different augmentation techniques: viz., Random Deletion, Synonym Replacement, Random Swap, Backtranslation, and NLPAug. The SPRAG corpus contains student responses involving keywords and special symbols. We experimented with four different data sizes with the augmented data to determine the impact of training data on the fine-tuned sentence transformer model. this paper provides an exhaustive analysis of fine-tuning pretrained sentence transformer models with varying sizes of data by applying text augmentation techniques. we found that applying random swap and synonym replacement techniques together while fine-tuning has given a significant improvement, With a 4.91% increase in accuracy and a 3.36% increase in the F1-score. All the trained models are publicly available11https://github.com/sridevibonthu/SPRAG/tree/main/augmentation..

K-sets and k-swaps algorithms for clustering sets

We present two new clustering algorithms called k-sets and k-swaps for data where each object is a set. First, we define the mean of the sets in a cluster, and the distance between a set and the mean. We then derive the k-sets algorithm from the principles of classical k-means so that it repeats the assignment and update steps until convergence. To the best of our knowledge, the proposed algorithm is the first k-means based algorithm for this kind of data. We adopt the idea also into random swap algorithm, which is a wrapper around the k-means that avoids local minima. This variant is called k-swaps. We show by experiments that this algorithm provides more accurate clustering results than k-medoids and other competitive methods.

Adapting K-Medoids and Random Swap for Clustering Sets

Adapting K-Medoids and Random Swap for Clustering Sets

Parallel random swap: An efficient and reliable clustering algorithm in Java

Solving large-scale clustering problems requires an efficient algorithm that can also be implemented in parallel. K-means would be suitable, but it can lead to an inaccurate clustering result. To overcome this problem, we present a parallel version of the random swap clustering algorithm. It combines the scalability of k-means with the high clustering accuracy of random swap. The algorithm is implemented in Java in two ways. The first implementation uses Java parallel streams and lambda expressions. The solution exploits a built-in multi-threaded organization capable of offering competitive speedup. The second implementation is achieved on top of the Theatre actor system which ensures better scalability and high-performance computing through fine-grain resource control. The two implementations are then applied to standard benchmark datasets, with a varying population size and distribution of managed records, dimensionality of data points and the number of clusters. The experimental results confirm that high-quality clustering can be obtained together with a very good execution efficiency. Our Java code is publicly available at: https://github.com/uef-machine-learning.

A Novel K-Means Clustering Method for Locating Urban Hotspots Based on Hybrid Heuristic Initialization

With rapid economic and demographic growth, traffic conditions in medium and large cities are becoming extremely congested. Numerous metropolitan management organizations hope to promote the coordination of traffic and urban development by formulating and improving traffic development strategies. The effectiveness of these solutions depends largely on an accurate assessment of the distribution of urban hotspots (centers of traffic activity). In recent years, many scholars have employed the K-Means clustering technique to identify urban hotspots, believing it to be efficient. K-means clustering is a sort of iterative clustering analysis. When the data dimensionality is large and the sample size is enormous, the K-Means clustering algorithm is sensitive to the initial clustering centers. To mitigate the problem, a hybrid heuristic “fuzzy system-particle swarm-genetic” algorithm, named FPSO-GAK, is employed to obtain better initial clustering centers for the K-Means clustering algorithm. The clustering results are evaluated and analyzed using three-cluster evaluation indexes (SC, SP and SSE) and two-cluster similarity indexes (CI and CSI). A taxi GPS dataset and a multi-source dataset were employed to test and validate the effectiveness of the proposed algorithm in comparison to the Random Swap clustering algorithm (RS), Genetic K-means algorithm (GAK), Particle Swarm Optimization (PSO) based K-Means, PSO based constraint K-Means, PSO based Weighted K-Means, PSO-GA based K-Means and K-Means++ algorithms. The comparison findings demonstrate that the proposed algorithm can achieve better clustering results, as well as successfully acquire urban hotspots.

Image Encryption Algorithm Based on Plane-Level Image Filtering and Discrete Logarithmic Transform

Image encryption is an effective way to protect image data. However, existing image encryption algorithms are still unable to strike a good balance between security and efficiency. To overcome the shortcomings of these algorithms, an image encryption algorithm based on plane-level image filtering and discrete logarithmic transformation (IEA-IF-DLT) is proposed. By utilizing the hash value more rationally, our proposed IEA-IF-DLT avoids the overhead caused by repeated generations of chaotic sequences and further improves the encryption efficiency through plane-level and three-dimensional (3D) encryption operations. Aiming at the problem that common modular addition and XOR operations are subject to differential attacks, IEA-IF-DLT additionally includes discrete logarithmic transformation to boost security. In IEA-IF-DLT, the plain image is first transformed into a 3D image, and then three rounds of plane-level permutation, plane-level pixel filtering, and 3D chaotic image superposition are performed. Next, after a discrete logarithmic transformation, a random pixel swapping is conducted to obtain the cipher image. To demonstrate the superiority of IEA-IF-DLT, we compared it with some state-of-the-art algorithms. The test and analysis results show that IEA-IF-DLT not only has better security performance, but also exhibits significant efficiency advantages.

A hybrid steady-state evolutionary algorithm using random swaps for Gaussian model-based clustering

A new hybrid evolutionary algorithm (EA) for Gaussian mixture model-based clustering is proposed. The EA is a steady-state method that, in each generation, selects two individuals from a population, creates two offspring using either mutation or crossover, and fine-tunes the offspring using the expectation maximization (EM) algorithm. The offspring compete with their parents for survival into the next generation. The approach proposed uses a random swap, which replaces a component mean with a randomly chosen feature vector as a mutation operator. In the crossover operator, a random component is copied from the source mixture into a destination mixture. Copying in crossover favors components of the source mixture located away from the components of the destination mixture. In computational experiments, the approach was compared to a multiple restarts EM algorithm, a random swap EM method, and a state-of-the-art hybrid evolutionary algorithm for Gaussian mixture model learning on one real and 29 synthetic datasets. The results indicate that, given the same computational budget, the proposed method usually learns mixtures with higher log-likelihoods than other benchmarks competing algorithms. The partitions of data obtained by the method correspond best to the original divisions of datasets into classes.

Random swapping, an effective and efficient way to boost the intensities of cross peaks in a 2D asynchronous spectrum

Analysis of mixture via chromatographic-spectroscopic and analogous experiments is a common task in analytical chemistry. A 2D/nD asynchronous spectrum is effective in retrieving spectra of pure substances even if different components cannot be separated. However, noise in the 2D/nD asynchronous spectrum becomes a bottleneck in the analysis. Finding a suitable sequence of the 1D spectra used in constructing the 2D/nD asynchronous spectrum is helpful to improve the signal-to-noise level. A 2D/nD asynchronous spectrum is often produced via a large number of 1D spectra. The resultant colossal number of the possible sequences makes stochastic search the only possible way to find a suitable sequence. Random changing (RC) and random swapping (RS) are two ways to obtain a new sequence. We found that the possibility of finding a better sequence via an RS is significantly higher than that via an RC in the advanced stage of stochastic searching. This is the reason why the performance of RS is superior to that of RC in two model systems where 2D asynchronous spectra are used. We applied the RS approach on the analysis of water/isopropanol mixtures, and satisfactory sequences are acquired with affordable computational cost. Thus, the RS approach brings about an opportunity increase the signal-to-noise level of a 2D asynchronous spectrum in the analysis of the bilinear data from complex mixed samples.

Computationally efficient topology optimization of scale-free IoT networks

The malicious attacks in the scale-free Internet of Things (IoT) networks create a serious threat for the functionality of nodes. During the malicious attacks, the removal of high degree nodes greatly affects the connectivity of the remaining nodes in the networks. Therefore, ensuring the maximum connectivity among the nodes is an important part of the topology optimization. A good scale-free network has the ability to maintain the functionality of the nodes even if some of them are removed from the network. Thus, designing a robust network to support the nodes’ functionality is the aim of topology optimization in the scale-free networks. Moreover, the computational complexity of an optimization process increases the cost of the network. Therefore, in this paper, the main objective is to reduce the computational cost of the network with the aim of constructing a robust network topology. Thus, four solutions are presented to reduce the computational cost of the network. First, a Smart Edge Swap Mechanism (SESM) is proposed to overcome the excessive randomness of the standard Random Edge Swap Mechanism (RESM). Second, a threshold based node removal method is introduced to reduce the operation of the edge swap mechanism when an objective function converges at a point. Third, multiple attacks are performed in the network to find the correlation between the measures, which are degree, betweenness and closeness centralities. Fourth, based on the third solution, a Heat Map Centrality (HMC) is used that finds the set of most important nodes from the network. The HMC damages the network by utilizing the information of two positively correlated measures. It helps to provide a good attack strategy for robust optimization. The simulation results demonstrate the efficacy of the proposed SESM mechanism. It outperforms the existing RESM mechanism by almost 4% better network robustness and 10% less number of swaps. Moreover, 64% removal of nodes helps to reduce the computational cost of the network.

Comparison of Data Augmentation Methods in Pointer—Generator Model

In this research, we proposed a data augmentation method using topic model for Pointer—Generator model. This method is that adding important sentences to an article as extended article. Furthermore, we compare our proposed method with data augmentation methods using Easy Data Augmentation (EDA), LexRank and Luhn. EDA consists of synonym replacement, random insertion, random swap, and random deletion. LexRank is based on Google’s search method and Luhn defines sentence features and ranks sentences. We considered which method is suitable for data augmentation. We confirm that most accurate model is the model using data augmentation method by topic model.