Search

Noise reduction is one of the main challenges for researchers. Classical image de-noising methods reduce the image noise but sometimes lose image quality and information, such as blurring the edges of the image. To solve this challenge, this work proposes two optimal filters based on a generalized Cauchy (GC) distribution and two different nature-inspired algorithms that preserve image information while decreasing the noise. The generalized Cauchy filter and the bilateral filter are two parameter-based filters that significantly remove image noise. Parameter-based filters require proper parameter selection to remove the noise and maintain the edge details. To this end, two filters are considered. In the previous works, the parameters of the mask that was made with the GC function were optimized and the mask size was considered fixed. By studying different noisy images, we find that the selected mask size significantly impacts the designed filter performance. Therefore in this paper, a mask is designed using the GC function to formulate the first filter, and despite the optimization of the filter parameters, the selected mask size is also optimized using the peak signal-to-noise ratio (PSNR) as a fitness function. In most metaheuristic-based bilateral filters, only the domain and range parameters, which are based on Gaussian distribution, are optimized and the neighboring radius is a constant value. Filter results on different noisy images show that the neighboring radius has a major effect on the filter performance. Since the filter designed with the GC function causes significant noise removal, this function is effective, and on the other hand, it’s almost similar behavior with the Gaussian function has caused it to be combined with the bilateral filter to design the second filter in this paper. The kernel of the domain and range is considered to be the GC function instead of the Gaussian function. The domain and range parameters and the neighboring radius are optimized using the PSNR as a fitness function. With the help of optimization algorithms such as the whale optimization algorithm and the Gaining sharing knowledge-based optimization algorithm, bilateral filter; and GC filter parameters are optimized. Finally, the performance of the proposed filters is investigated on images corrupted by Gaussian and impulse noise. It is compared with other classical filters, the particle swarm optimization (PSO) based GC filter, and two PSO-based bilateral filters on various images. The experimental findings demonstrate that the suggested filters outperform the others.

Many papers in the intersection of theoretical and applied algorithms show that the simple, asymptotically less efficient algorithm, performs better than the bestcomplex theoretical algorithms on random data or in specialized “real world” applications. This paper considers the Knuth–Morris–Pratt automaton, and shows a counter-intuitive practical result. The classical pattern matching paradigm is that of seeking occurrences of one string—the pattern, in another—the text, where both strings are drawn from an alphabet set Σ. Assuming the text length is nand the pattern length is m, this problem can naively be solved in time O(nm). In Knuth, Morris and Pratt’s seminal paper of 1977, an automaton, was developed that allows solving this problem in time O(n) for any alphabet. This automaton, which we will refer to as the KMP-automaton, has proven useful in solving many other problems. A notable example is the parameterized pattern matchingmodel. In this model, a consistent renaming of symbols from Σis allowed in a match. The parameterized matching paradigm has proven useful in problems in software engineering, computer vision, and other applications. It has long been believed that for texts where the symbols are uniformly random, the naive algorithm will perform as well as the KMP algorithm. In this paper, we examine the practical efficiency of the KMP algorithm versus the naive algorithm on a randomly generated text. We analyze the time under various parameters, such as alphabet size, pattern length, and the distribution of pattern occurrences in the text. We do this for both the original exact matching problem and parameterized matching. While the folklore wisdom is vindicated by these findings for the exact matching case, surprisingly, the KMP algorithm alwaysworks significantly faster than the naive in the parameterized matching case. We check this hypothesis for DNA texts and image data and observe a similar behavior as in the random text. We also show a very structured exact matching case where the automaton is much more efficient.

Cardiovascular disease (CVD) is one of the most contributing diseases to premature mortality around the globe. Low- and middle-income countries like India account for almost 80% of global CVD fatalities. Predicting cardiovascular diseases at an early stage can improve the quality of life. Electrocardiography (ECG) is one of the non-invasive methods to assess heart function disorders and CVDs. The paper presents the prediction of conduction disturbance or disorders (CD) through a 12-lead electrocardiogram (ECG) that leads to chronic heart failure or cardiac arrest. Three ensemble machine learning models i.e., random forest (RF), XGBoost, and the support vector machine, are used to classify the conduction disturbance subjects from the ‘normal’ subjects. In addition to this, the paper also presents a comparative study to show the effect of two demographic features, ‘age’ and ‘sex’ on the prediction of conduction disturbance’s subjects. The performance of the classifiers’ is measured in terms of accuracy, precision, recall, and F1 score. Tenfold cross-validation is utilised, and the receiver operating curve is traced for each of the combinations of tenfold cross-validation. The performance is measured with a confusion matrix for all three classifiers. The performance with RF and XGBoost performance is similar in terms of accuracy, whereas the total number of true predictions is higher in the case of RF. The proposed model would be useful for continuous monitoring and prediction conduction disturbance in the smart healthcare framework.

Successful cyber attacks against industrial systems, such as water treatment systems, can lead to irreparable consequences for public health and the economy. Machine learning and deep learning could help detecting and forecasting previously unknown cyber attacks but require specific datasets. The number of publicly available datasets in this field is very limited and the majority of the publicly available datasets used in cyber security tasks have severe flows. In this paper, the authors introduce the unified methodology for the generation of the dataset for industrial water treatment security. Detailed specification of stages of the methodology is given. The paper ends with a usage scenario describing preparatory stages for dataset generation for the cybersecurity research in water treatment systems, namely, specification of the technological process, testbed development, and development of the attack model for the considered technological process. The developed methodology will be used for the dataset generation, that, in turn, will be used to develop and test cyber attack detection methods based on machine learning and deep learning, and to strengthen the water treatment systems’ security.

Large language models (LLMs) have garnered significant attention lately, and one particular implementation that has captivated users is ChatGPT, a first-of-its kind innovation that sparks intense debates among professionals regarding its ethical boundaries, especially after it was released for public use. Some of the debates are ongoing, with several people getting more used to this tool and some losing touch with several ethical concerns. Thus, it is crucial to explore the early professional’s perspectives surrounding ChatGPT to comprehend its potential impact on the educational sector and, in particular, society in general. This research aims to address this knowledge gap by employing data mining techniques on LinkedIn to analyze ChatGPT-related discussions and provide a comprehensive textual examination. The dataset for this investigation consists of 166 records written in English by LinkedIn users discussing ChatGPT in its early stage. The study utilized VOSviewer software to clean, analyze, mine, and visualize the data. Among the sample, it was observed that 49% of the users are PhD holders, predominantly with experience in scientific research and data science fields. Hence, the findings reveal that issues such as plagiarism, references, citations, manuscripts, papers, and literature reviews stand out as the major concerns. These results suggest that while ChatGPT can contribute to academic progress by fostering new knowledge, thoughts, and ideas, it can also facilitate academic misconduct, including plagiarism and inaccurate information dissemination. Therefore, this paper also provides discussions on the implications and ethical issues of ChatGPT. In particular, it is imperative to establish clear ethical boundaries and guidelines for the appropriate use of LLMs in academic settings.

The human body responds to challenging events or demanding conditions by entering an escalated psycho-physiological state known as stress. Psychological stress can be captured through physiological signals (ECG, EEG, EDA, GSR, etc.) and behavioral signals (eye gaze, head movement, pupil diameter, etc.) Multiple stressors occurring simultaneously can cause adverse mental and physical health effects, leading to chronic health issues. Continuous monitoring of stress using wearable devices promises real-time data collection, enabling early detection and prevention of stress-related issues. The present study presents an extensive overview of stress detection techniques utilizing wearable sensors, and machine & deep learning methods. Various types of machine learning algorithms, such as logistic regression, support vector machines, random forest, and deep learning models, have been applied to analyze data and detect stress. Additionally, the paper highlights the stress detection datasets and research conducted in different areas. The challenges and obstacles related to detecting stress using machine intelligence are also addressed in the paper.

Most parallel scientific programs contain compiler directives (pragmas) such as those from OpenMP (Hermanns in Parallel programming in Fortran 95 using openMP, 2002. School of Aeronautical Engineering, Universidad Politécnica de Madrid, España, 2011), explicit calls to runtime library procedures such as those implementing the Message Passing Interface (MPI) (in A message-passing interface standard version 4.0, 2021. https://www.mpi-forum.org/docs/mpi-4.0/mpi40-report.pdf), or compiler-specific language extensions such as those provided by CUDA (Ruetsch and Fatica in CUDA Fortran for scientists and engineers: best practices for efficient CUDA Fortran programming, Elsevier, 2013). By contrast, the recent Fortran standards empower developers to express parallel algorithms without directly referencing lower-level parallel programming models (Numrich in Parallel programming with co-arrays, CRC Press, 2018, and Curcic in Modern Fortran: building efficient parallel applications, Manning Publications, 2020). Fortran’s parallel features place the language within the Partitioned Global Address Space (PGAS) class of programming models. When writing programs that exploit data parallelism, application developers often find it straightforward to develop custom parallel algorithms. Problems involving complex, heterogeneous, staged calculations, however, pose much greater challenges. Such applications require careful coordination of tasks in a manner that respects dependencies prescribed by a directed acyclic graph. When rolling one’s own solution proves difficult, extending a customizable framework becomes attractive. The paper presents the design, implementation, and use of the Framework for Extensible Asynchronous Task Scheduling (FEATS), which we believe to be the first task scheduling tool written in modern Fortran. We describe the benefits and compromises associated with choosing Fortran as the implementation language, and we propose ways in which future Fortran standards can best support the use case in this paper.

This paper is an in the wild evaluation of the eye tracking tool Pistol. Pistol supports Pupil Invisible projects and other eye trackers (Dikablis, Emke GmbH, Look, Pupil, and many more) in offline mode. For all eye tracking recordings, Pistol is able to compute pupil and iris ellipse, eye aperture, eye lids, 3D eye ball, optical vector estimation, and eye movement types. For Pupil Invisible projects, it also supports the detection of a special marker and allows the 2D and 3D gaze estimation mapped on the scene camera. In this extended journal paper for Pistol we focus on challenging in the wild situations, such as reflections on glasses, eye tracking during sport, changing lightning conditions, and this for two eye trackers. The tool Pistol can be downloaded here Link

This paper provides a solution to mitigate mistrust that arises from untransparency in a multi-stakeholder environment. This work is related to a kind of verifiable computation. When considering the multi-stakeholder environment, not only the participants’ requirements but also the assignee’s intention should be respected. That is, the assignee should be given the discretion to select a result that is the best for the assignee among all the choices. However, there is a possibility that if the assignee is malicious, he/she may falsify and provide an inadequate result for participants to maximize his/her benefit by ignoring the participants’ requirements. It is difficult for the participants to detect this if they want to keep their preference orders secret from others. This paper proposes a solution to determine whether the received result is adequate for the participants while keeping their preference orders secret. The proposed solution is based on fully homomorphic encryption (FHE) and assumes the use of a semi-honest third-party server. This paper first describes a general solution that is not limited to specific requirements from participants. Next, this paper shows a way to apply to stable matching problem as a specific implementation. More specifically, a transformation of a standard stability checking algorithm into an algorithm that can be implemented by FHE with the computational complexity O(n2). Finally, this paper gives an example of an implementation and its performance with HElib, which is an FHE library that provides BGV.

Botnet detection systems are becoming more important as cybercriminals continue to develop new Bot tools and applications. A botnet is a collection of several compromised systems that are connected to the central controller called a botmaster. These compromised devices are carried out various malicious activities, such as DDoS attacks, phishing, Email Spam, identity theft, stealing personal credentials of the user, etc. due to the dynamic change in the botnet size, it is difficult to detect the botnet. As long as botmasters are coming up with new ways to attack, sophisticated solutions for botnet detection are very essential. To illustrate how to use these tools, this paper will discuss several tools and processes involved in developing a Botnet detection system. Different libraries like Scikit Learn, Pandas, Theano, Matplotlib, Pickel, and NumPy are used. Additionally, the processes for utilising these tools are illustrated in this paper. The features are extracted like packet size, packet byes, source address, destination address, length, and corresponding protocols. Feature extraction requires a significant amount of domain expertise and manual work from professionals in current machine learning-based botnet detection systems. Botnets are divided based on their protocol, such as Internet relay chat, DNS, and P2P which are used by the C&C Server. In this paper, we proposed a model to detect Botnet using three machine learning algorithms, i.e., K-Nearest Neighbor (KNN), Decision Tree (DT), and Naive Bayes (NB) for the experiments on a dataset among these three classifiers NB performs the best and has an accuracy of 90.62%.