Automated brain tumor classification has garnered considerable attention as a research area, with Convolutional Neural Networks (CNN) and deep learning emerging as the standard tools for developing cutting-edge solutions. In this paper, we propose a Vision Transformer (VIT) architecture and examine their efficacy and limitations. Our evaluation process encompasses various parameter configurations, including different image size, transformer block number, and patch size. We conduct training and testing on a publicly available brain tumor classification dataset containing 3064 images across three tumor classes: meningioma, glioma, and pituitary tumor. Through rigorous evaluation, we find that our proposed VIT model can achieve competitive performance compared our previous CNN model. The best results include an overall Dice similarity score and correct decision rate of 98.9%, with AUC values exceeding 99.6% for each tumor class.
Every year, the number of organizations affected by ransomware attacks increases exponentially, with over five thousand incidents in 2024 alone. This trend shows no signs of slowing down, despite the best efforts of security companies, researchers and government agencies. Ransomware - the practice of encrypting compromised systems and demanding a ransom payment for their release - remains the most prolific form of cybercrime to date. There are still countless Internet-exposed systems running outdated or improperly configured software, and the healthcare industry is one of the most affected sectors worldwide. Healthcare institutions often lack sufficient funding and technical expertise to properly secure their systems, which are both crucial for day-to-day operations and contain confidential patient information. These factors make them prime targets for malicious actors, because disabling these systems can severely impact patient care, and without sufficient business continuity planning and technical staff, hospitals can become crippled by cyberattacks. The goal of this paper is to demonstrate the severity of ransomware attacks targeting healthcare institutions, by highlighting notable case studies and research documenting the various problems that cyberattacks can cause in a hospital. Based on the latest statistics and trends published by cybersecurity companies, I draw the conclusion that underfunded and vulnerable systems in the health sector are a major contributing factor in the rise of Ransomware-as-aService groups, whose continuous assaults negatively impact patient health and safety.
This paper aims to explore the feasibility of conducting Health Technology Assessment analysis using published costs transferred to the target country. It evaluates four strategies for imputing missing costs for type 2 diabetes and seven of its possible complications using the UK Prospective Diabetes Study (UKPDS) model. Our goal is to determine the cost transfer strategy that provides the most reliable and consistent cost imputation while strictly following the predefined cost transfer formula. The study compares four strategies with varying levels of flexibility in accepting transferred costs from proxy diseases and distant countries. Results indicate that more flexible strategies offer a more comprehensive approach for transferring costs when local data are unavailable, although outcomes may vary depending on the chosen scenario. Moving forward, we will apply this formula to both strategies, compare the resulting estimates to true costs from Saudi Arabia, and draw conclusions on the accuracy and reliability of each strategy's cost estimation.
Research on microplastics (MP), plastic particles smaller than 5mm, has recently gained significant attention due to their environmental and health risks. Research focuses on identifying MP sources, their distribution in the environment, and their impact on organisms and human health. This systematic review aims to summarize research on the health effects of MPs on humans, detailing specific health impacts linked to different types of microplastics. A systematic search in the PubMed database up to December 25, 2024, and a review of previously published studies in the field identified only 4 systematic reviews that have been published on the human health effects of microplastics. We summarized the health effects of MPs reported in the identified studies and their specific health research focus. The PubMed database search yielded 8,018 hits on microplastic, 82 records provided systematic review, and 29 full text papers were considered eligible for the studies of health effect. Of the 29 full-text papers 4 reported on the human health effects of microplastics. Given the health impacts of microplastics on humans, the research highlights areas that need further study. This is particularly important as the health effects of microplastics are not yet fully understood, and many questions remain unanswered.
Our research focuses on generational management, which requires a comprehensive understanding of the generational characteristics of employees present in the labour market. At the beginning of our research, we started from this premise and analyzed the key characteristics of the generations currently active in the labour market. In the modern workplace, multiple generations work together, each driven by different motivations and influences, shaped by the unique historical and social contexts they have experienced. As a leader, it is often challenging to understand the motivations and value systems of different employees. Throughout our research, we examined the main characteristics of generational differences, their interrelations, and their core values. We highlighted the importance of leaders recognizing these differences in order to effectively support and motivate employees, ensuring that different generations are wellmanaged using appropriate leadership techniques and tools. To achieve this, we analyzed leaders from different generations. We then compared our findings and proposed a generational shift in leadership approach, supported by a leadership model.
This study evaluates the effects of negative productivity shocks on consumption, and positive government spending shocks on productivity using a simple VAR model. Secular stagnation is not a new concept in economics. Milton Friedman predicted the stagflation of the 1970s in the U.S. using the idea of a long-term Phillips curve, stating the independence of inflation and unemployment in the long run. Blanchard and Summers investigated the stagnating employment growth of Europe and came to the conclusion that a key driver of secular stagnation is demand shocks that permanently alter potential output from its previous course. As the global economy exited the Great Moderation through the Great Financial Crisis of 2008, a jobless recovery ensued bringing the concept of hysteresis in the spotlight. The sharp COVID downturn brought an end to 2010s' age of austerity, and governments around the world implemented some forms of a high-pressure economy to encourage consumption and avoid an economic depression. Through an impulse response function, based on a model leveraging historical time series relationships, we find evidence of permanent output-altering effects of negative productivity shocks.
This paper explores various methods for assessing and mitigating urban heat islands (UHI), including observational techniques, numerical modeling, and satellite-based measurements. IDA ICE simulations provide insights into building-scale thermal behaviors, while mesoscale and microscale models facilitate detailed analyses of urban thermal dynamics. Satellite observations, particularly MODIS and Landsat, offer large-scale data but are affected by cloud cover and seasonal variations. Direct measurement methods, such as meteorological stations and mobile sensing, complement these modeling techniques by validating results. Integrating these varied methodologies allows for a comprehensive understanding of UHI phenomena. This study underscores the significance of interdisciplinary approaches and technological advancements in addressing UHI and promoting sustainable urban development.
Frequent meat-safety-related incidents have increased concern about meat safety and quality among both consumers and the entire meat industry. Given that improving traceability in meat supply chains has been a crucial aspect of ongoing efforts to enhance meat safety and reduce the risk of contamination, it is essential to understand consumer perceptions about traceable meat. This study employs a qualitative research method, applying focus group interviews with twenty consumers to explore their perceptions of meat safety, the factors influencing their purchase and consumption behaviors, and their perceptions about traceable meat products, in particular. The study found that consumers are strongly concerned about meat safety, and purchasing behavior is influenced by sales channels, the characteristics of meat, and social media. While most participants accepted the credibility of traceable meat, some consumers expressed a lack of trust in the self-regulation of the industry. The research results are expected to contribute to the effective development of the local livestock industry and guide the meat industry in introducing and developing a meat traceability system, thereby restoring consumer trust.
The rise of global environmental issues and the growing emphasis on sustainability necessitate a transformation of economic sectors. Producers must adapt to evolving social values and consumer demands. The first step toward sustainable agricultural production is chemical-free organic farming, a practice that positively impacts both ecosystem protection and human health. These aspects are particularly important to environmentally conscious and health-aware consumers, who consider not only product quality but also the production methods behind it when making purchasing decisions. In our study, we provide an overview of the global landscape of organic farming, and compare the findings of two previous primary studies, - one examining the current state and development opportunities from the producers ' perspective the other from that of the consumers. Based on this comparison, we formulate proposals for the development opportunities of organic farming.
Analyzing the financial and environmental impacts of technological developments in waste management is crucial for economic sustainability. This paper presents a soft computing method for optimizing performance and consumption in decision-making. Waste management equipment requires performance and consumption data. Waste composition varies by region and season, affecting shredder performance. Soft computing techniques, such as fuzzy logic, manage changing conditions and optimize equipment use. The Mamdani system helps select appropriate equipment and plan costs by handling performance and consumption uncertainty. The Mamdani system was used in the Fuzzy Logic Toolbox of MATLAB for design and simulation using Fuzzy Logic Designer, defining “Consumption” and “Performance” in“Low” “Medium”, and “High” fuzzy sets. After defining rules and defuzzification with Centroid method, the developed Mamdani surface showed compliance values with “low” (38 liters/hour) consumption and “high” (50 tons/hour) performance, resulting in a “good” performance output of 0.163. In summary, a new fuzzy logic model was developed, offering significant advantages in green waste management by handling uncertain conditions, indirectly contributing to increased collection and use of green waste.
This article discusses optimization possibilities for a real-time attacker profiling system that makes use of honeypots. We revisit an already implemented system that aimed to classify attackers according to the patterns of observed attacks in honeypot environments. The architecture of the implemented solution combines the collection of data from deployed honeypots with a classification system to categorize attackers in real time. The chapter discusses the following key components of the system: honeypot network design, central log management through the ELK stack, including Elasticsearch, Logstash, Kibana, and a custom-developed data processing module. We present the implemented methodology for the identification and scoring of various attack patterns, from simple port scans to more sophisticated intrusion attempts. Optimization possibilities were investigated in order to enhance the capability of the proposed system for rapid and effective profiling of cyber attackers. These involve an assessment of the attack detection algorithms, the scoring system, and the refinement of a more complete attacker classification model. The paper concludes with the testing results in both a controlled Capture The Flag event and an actual internetexposed deployment, thus showing results on the performance of the system under different conditions. By proposing such optimizations, we want to improve the effectiveness of the system in promptly providing information about potential threats and allowing better and more focused response strategies against them in current cybersecurity operations.
Deep Q-Networks have limited usability in multiagent scenarios. As older memories of the experience replay buffer tend to cause problems with learning, usually the length of the experience replay buffer is shortened. In this paper, a new method is proposed with which the buffer length can be kept greater. This is performed by storing the Q -values in the buffer, and upon training, the error of the softmax of the current model and the softmax of the past Q-values are taken as a weight for learning. This method is then benchmarked on environments where the agents have to keep as far from each other as possible. The two-dimensional environment can be correlated to a real-world engineering problem of radio transmission coverage.
There are numerous papers in the venture capital topic. Each of them evaluates certain aspects of venture capital investments. Relatively only a small subset deals with investment strategies. Most of the related papers are only partly related and generally are not aimed to address investment strategy matters. There is an increasing interest towards venture capital investments hence questions related investment effectiveness are equally important. We believe there is a potential research gap in this area because how to run a profitable venture capital fund is a very complex question. Moreover, it is believed it is very hard to study it. Consequently, this paper aims to detail a scoping review research protocol for venture capital investment strategies. We try to formulate a detailed substructure of the strategy domain to help to map this important subdomain.
This paper is about new findings in a recent research program to enhance automation of research mainly in high industrial automation related topics applying All-in-One experimental model of research to represent and integrate component models, simulations, and context driven communications. Reactively integrated model mediated research is aimed to realize. Reactive integration in this model means the capability to react for relevant inside outside object parameter changes by analysis then autonomous operations to update research model by new or modified model objects and contextual connections. Step-by step and trackable development of research model is considered which is realized using relevant model definition and generation activities in a research eligible and suitable engineering modeling, simulation, and collaboration platform. Context changes generate driving attempts which are autonomously processed by executable research models whereas measures are applied at avoiding erroneous recognition, decision, and driving. Capabilities for research integration are extended to an innovation cycle and are surveyed in this paper.
This paper presents an innovative calibration technique for low-cost MEMS accelerometers, using an Adaptive Neuro-Fuzzy Inference System (ANFIS) optimized by a Genetic Algorithm (GA). The proposed method aims to enhance accuracy in inertial measurement units (IMUs), which are prone to errors such as biases, misalignments, and scalefactor variances. Employing a UR5 robotic arm for performing the motions for the calibration enabled gathering the groundtruth information simultaneously with the sensor readings. The method calibrates IMU data in 3D space using dynamic motions, eliminating the need for high-cost rotation rigs. Separate ANFIS models were trained for each sensor axis using both triaxial accelerometer data and Euler angles, achieving significant error reductions compared to the traditional ellipsoid fitting method. Improvement of 11.34 % could be gained using the ANFIS models compared to the ellipsoid fitting. Experimental results confirm that the proposed ANFIS-GA method effectively compensates deterministic and stochastic errors, enhancing sensor reliability for applications in navigation and motion tracking.
Food safety is important for everyone's daily life, which is also connected to food security, but it is not perceived well by people, and this topic is under-researched. Awareness of food safety significantly influences public health, national policies, and individual behavior. However, the current research is distributed unevenly in food safety awareness among the main players across the food value chain, such as farmers, transporters, processors, food service staff, and consumers. Identifying the food safety awareness level across the food value chain comprehensively and the characteristics are critical to policymakers, organizations, and individuals to have effective strategies to improve or strengthen food safety awareness and practices. In this primary research, we employed a computational intelligence approach, cluster analysis, to distinguish the food safety awareness patterns. The statistical analysis results from 328 sampling populations in China revealed two distinct groups, the “food safety-conscious” and “food safetyunaware” groups. People who have relatively good food safety awareness tend to have better food safety practices and are categorized as “food safety-conscious” and vice versa for the “food safety-unaware” group. People's sociodemographic characteristics tend to have an impact on food safety awareness and practices, but not at a significant level, such as age, living area, occupation status, monthly income, and highest education.
Accurate localization is essential for wheeled robots in structured environments. This study evaluates the performance of an EKF-based sensor fusion approach that combines odometry with an Absolute Positioning System (APS) under different configurations, analyzing how APS setup variations affect localization accuracy. A line-following, differentially driven wheeled robot was used, providing odometry data. The Ground Truth (GT) database consists of predefined, precisely measured trajectories on which the robot was deployed. Based on these trajectories and odometry readings, APS data were artificially generated, providing virtual measurements with different noise levels. Five trajectories were analyzed, including multiple straight-line segments with directional changes and one semi-circular path. The APS was tested at two update frequencies (1 Hz and 5 Hz) and five noise levels (20,40,60,80, and 100 mm standard deviation (STD)). Performance was evaluated using Root Mean Square Error (RMSE), STD, and Maximum Absolute Error (MaxAE). Results show that EKF-based fusion improves accuracy on straight trajectories with sudden direction changes, while for the semicircular path, odometry alone performs comparably. Higher APS update frequencies enhance localization, while increased noise degrades it. To achieve RMSE lower than 50 mm, the noise level must not exceed 20 mm for 1 Hz APS and 40 mm for 5 Hz APS.
Organizational problems can take various forms. Fundamentally, they may relate to people and processes. On the process side, key activities include production, procurement and logistics, quality assurance, marketing and sales support, customer experience, as well as leadership, finance, and other processes that contribute to efficient and effective operations. The human aspect is more complex and nuanced, as it involves a highly intricate and interdependent interaction with organizational culture. In our research, we examined the challenges-primarily related to organizational development-that Hungarian business leaders face in managing their companies. The most frequently reported problems were: Challenges in retaining employees and the difficulties associated with workforce replacement; Deficiencies in internal communication and disruptions in the flow of information. Ineffectiveness of leadership structures and delays in decision-making processes; Lack of recognition and the challenges of maintaining employee motivation. Monthly frequented issues, in order, are as follows: excessive meetings and discussions without prior planning; Inadequate communication within the organization; Disorganization in some processes, leading to decreased efficiency; Indecisiveness; Lack or difficulty in workforce succession (recruiting); Lack of recognition, Lack of control over delegated tasks, Incomplete or meaningless task assignments, lack of awareness of each other's work. Recruitment difficulties remain a persistent challenge for organizations, and conflict management is the most widely applied problem-solving method.
Anatomical magnetic resonance images are affected by different types of noise, including thermal, motion, radio interference, and magnetic field inhomogeneities. In a clinical setting, acquiring MR images of the highest quality is not always feasible. Quantitative and artificial intelligence-based decision support tools require high-quality data to accurately differentiate among pathological conditions, avoiding diminishing the clinical relevance of a diagnostic model. A fully convolutional model with no pooling layers was trained on a set of noisy images, with the ground truth being the original image without the noise. Different levels of noise were incorporated into the training set. The experiments showed a reduction in noise levels, but it can impact quantification tasks when T2ws without noise are provided to the model. Six types of pairs of original T2w image slices and the corresponding slices with synthetic noise were generated with various thresholds of Gaussian noise, spanning from 4% to 14%. In total, 38500 pairs were utilized for convergence and evaluation of the proposed denoising models. The examined deep denoiser reduced improved image quality by up to 18.2% peak signal-to-noise ratio (PSNR), overall across the aforementioned noise thresholds.
