Journal of Soft Computing and Computational Intelligence (e-ISSN: 3048-6610)

Ethics of Soft Computing in Decision-Making

Sanjay Kumar — Wed, 06 Nov 2024 00:00:00 +0000

Soft computing has revolutionized decision-making processes across various domains, including healthcare, finance, and autonomous systems, by providing flexible and adaptive solutions to complex problems. However, the increasing reliance on these techniques raises significant ethical concerns that must be addressed to ensure responsible implementation. This article explores critical ethical issues surrounding the use of soft computing in decision-making, such as accountability, bias, transparency, privacy, and the broader societal impacts of automated decisions. As algorithms take on more responsibility in critical areas, questions about who is accountable for their outcomes become paramount. Additionally, the potential for bias in algorithmic decision-making can lead to unfair treatment of individuals and groups, necessitating effective mitigation strategies. Transparency and explainability are crucial to fostering trust, as many models operate as "black boxes," complicating the understanding of their decision processes. Privacy concerns also emerge due to the extensive data collection often required by these systems. Finally, the societal implications, including job displacement and erosion of human agency, warrant careful consideration. By analyzing these issues, this article proposes a framework for ethical concerns that can guide the development and application of soft computing technologies, ensuring they align with societal values and promote equitable outcomes.

Advanced Aerial Mobility: Innovative Urban Air Transportation System

Aarth Anant Dahale, Prashant Lahane — Fri, 08 Nov 2024 00:00:00 +0000

Recent advancements in computer vision, deep learning, and GPS technologies have enabled the creation of sophisticated flying drone transportation systems. These systems use advanced algorithms and sensors to navigate safely and efficiently, potentially transforming transportation through autonomous delivery, monitoring, and emergency response. This paper reviews the application of YOLO (You Only Look Once) and R-CNN (Region-based Convolutional Neural Network) algorithms for object detection and tracking in drones, providing them with enhanced awareness and responsiveness to dynamic environments. Furthermore, Deep Neural Networks (DNNs) support autonomous navigation and decision-making, enabling drones to maneuver complex routes and make real-time adjustments in response to unforeseen obstacles.

The literature review offers an in-depth analysis of these technologies, examining how they collectively enhance drone capabilities. The paper also discusses the barriers that must be addressed to facilitate the integration of flying drones into current transportation systems. These challenges include regulatory considerations, airspace management, safety protocols, and public concerns regarding privacy and security. Addressing these issues is essential for drones' safe, ethical, and sustainable adoption in urban settings. Overall, this review underscores the transformative potential of flying drone technology while highlighting the importance of a coordinated approach to ensure responsible deployment.

A Comprehensive Study Robust Statistical Methods for Detection of Malware

P. Devi Sravanthi, Manas Kumar Yogi — Fri, 08 Nov 2024 00:00:00 +0000

This paper explores the application of robust statistical methods for detecting malware, addressing the challenges posed by noisy data and evolving threats in cybersecurity. Traditional malware detection techniques often rely on fixed signatures, making them vulnerable to new variants. In contrast, robust statistical methods, such as outlier detection and robust regression, effectively identify strange patterns in network traffic and system behavior, enabling the recognition of previously unseen malware. We discuss integrating these methods with machine learning algorithms to enhance detection accuracy and resilience. Techniques like kernel density estimation help establish baseline behavior, facilitating the identification of deviations indicative of malicious activity.

Additionally, Bayesian approaches allow for dynamic model updates, providing real-time adaptability to new data. Our findings demonstrate that robust statistical methods significantly improve the reliability of malware detection systems, particularly against sophisticated attacks and zero-day exploits. By leveraging these advanced techniques, organizations can enhance their cybersecurity posture, effectively mitigating risks associated with evolving malware threats. This paper highlights the importance of incorporating robust statistical approaches into malware detection frameworks to achieve greater accuracy and resilience in an increasingly complex cyber landscape.

A Comparative Study of Deep Learning Models in Detection of Lung Opacity

G. Keerthi, G. V. S. N. R. V. Prasad, M. Babu Rao — Thu, 28 Nov 2024 00:00:00 +0000

Lung opacity detection is vital in diagnosing various pulmonary conditions, such as pneumonia, tuberculosis, and COVID-19. This study examines the efficacy of three deep learning models Visual Geometry Group 16 (VGG16), ResNet50, and Dense Convolutional Network121 (DenseNet121) in identifying lung opacity from chest X-ray images. Using datasets sourced from the COVID-19 radiography dataset and Kaggle, these models were trained and evaluated to determine their performance. The data underwent preprocessing to improve model accuracy and generalizability, including resizing, normalization, and augmentation. Each model was assessed based on classification accuracy, demonstrating distinct levels of effectiveness. DenseNet121 outperformed its counterparts with an accuracy of 91.90%, attributed to its dense connectivity that enhances gradient flow and feature reuse. VGG16 achieved an accuracy of 89.99%, benefiting from its simplicity and structured design. ResNet50, despite its more profound architecture and skip connections, lagged with an accuracy of 75.79%, likely due to challenges in capturing subtle patterns within the data. The results establish DenseNet121 as the most suitable model for lung opacity detection, offering a reliable and efficient solution for clinical applications. Its high accuracy suggests potential integration into diagnostic workflows, where quick and precise detection is critical for improving patient outcomes. Future work includes utilizing larger datasets, implementing ensemble techniques, and integrating explainable deep learning (DL) methods to improve model interpretability and reliability, promoting adopting deep learning solutions in medical imaging.

Flirting Words Detection Using Machine Learning Techniques

Alugolu Avinash, Pamulapati Lakshmi Satya , Gaduthuri Alekhya — Thu, 19 Dec 2024 00:00:00 +0000

Flirting is a form of interpersonal communication often involving subtle language cues, making it challenging to identify accurately. Detecting flirting words or patterns can be valuable in various applications, such as social media moderation, dating platforms, and conversational AI systems. This study explores machine learning techniques to detect flirting words or phrases in textual communication. The research creates a labeled dataset on flirtatious and non-flirtatious text samples, where NLP will be applied to preprocess and then feature-engineer the data. Multiple machine learning models, including supervised and unsupervised algorithms, are compared using their accuracy in flirting cue identification. The study also compares the model's accuracy enhanced by semantic understanding, the analysis of context, and sentiment detection. The results show that state-of-the-art models such as deep learning and transformer-based architectures like BERT have outperformed classical machine learning techniques to capture subtle aspects of flirting language. Thus, results highlight the role of contextual and cultural factors for conversational AI in identifying flirtation and underscore the future directions of enhancing machine learning that would be able to comprehend and emulate human conversation better.