The rapid expansion of devices on the Internet of Things (IoTs) has led to a significant rise in IoT botnet attacks, creating an urgent need for advanced detection and classification methods. This study aims to evaluate the effectiveness of Kolmogorov-Arnold Networks (KANs) and their architectural variations in classifying IoT botnet attacks, comparing their performance with traditional machine learning and deep learning models. We conducted a comparative analysis of five KAN architectures, including Original-KAN, Fast-KAN, Jacobi-KAN, Deep-KAN, and Chebyshev-KAN, against models like Multi-Layer Perceptron (MLP), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRU). The evaluation was performed on three IoT botnet datasets: N-BaIoT, IoT23, and IoT-BotNet, using metrics such as accuracy, precision, recall, F1-score, training time, and model complexity. KAN variants consistently demonstrated robust performance, often exceeding traditional ML and DL models in accuracy and stability across all datasets. The Original-KAN variant, in particular, excelled in capturing complex, non-linear patterns inherent in IoT botnet traffic, achieving higher accuracy and faster convergence rates. Variations such as Fast-KAN and Deep-KAN offered favorable trade-offs between computational efficiency and modeling capacity, making them suitable for real-time and resource-constrained IoT environments. Kolmogorov-Arnold Networks prove to be highly effective for IoT botnet classification, outperforming conventional models and offering significant advantages in adaptability and accuracy. The integration of KAN-based models into existing cybersecurity frameworks can enhance the detection and mitigation of sophisticated botnet threats, thus contributing to more resilient and secure IoT ecosystems.

This paper discusses feature extraction methods. The feature extraction methods such principal component analysis and multiple discriminant analysis are very important techniques in machine learning research areas. The characteristic of feature extraction is to transform the data from a difficultly classified space to a easily classified space. There are many conventional machine learning methods including transformation such as artificial neural network and support vector machines. However, extracting the good features before applying machine learning methods will lead to better classification results. This paper focuses on the principal component regression (PCR). The PCR finds the approximation with hyper-planes where the data distributed on. The problem now is that it has a case of the data do not distribute on hyper-planes, for example they distribute on hyper-spheres such as rotation objects, the PCR can not extract the good feature to apply the classification problems. This paper proposes a new feature extraction method by calculating the conformal eigenvectors in conformal geometric algebra(CGA) space to find the approximation hyper-planes or hyper-spheres which fit to the set of data using the least square approach. In particular, this paper shows that the classification accuracy of proposed method is better than that of conventional PCR method.

Numerous attacks on the Internet of Things (IoTs) necessitate detection and prevention technologies, such as Intrusion Detection Systems (IDSs). Anomaly detection is a time-consuming and challenging task in intrusion detection systems. It requires the development of a robust classifier model capable of intelligently detecting multiple types of attacks. When different input features are provided, IDS' detection performance will fluctuate substantially. Additionally, the volume of network traffic and its multidimensional characteristics will result in a lengthy classification process. In this paper, we focus on data preprocessing and propose another feature extraction method for classifying attacks in IoT networks. Additionally, we also point out the features that have the most significant impact on the accuracy of the classification model for malicious IoT traffics.

Human Activity Recognition is one of the most important areas of computer vision research. Its applications include intelligent security monitoring systems, health care systems, intelligent transportation systems, and a variety of systems that involve interactions between people and electronic devices such as human computer interfaces. However, the biggest difficulty for human activity recognition system is data collected from the camera can be deficient or analytical methods are not capable of performing the entire distribution of data when camera moves. Further more, low recognition results also is a particular obstacle. In this study, we present a new method which is a marker selection for human activity recognition using combination of conformal geometric algebra and principal component regression. First, from the C3D motion data we used marker selection method to select the best markers. Then, using CGA to represent the motion of 3D objects. Next, PCR is used to reduce the dimensionality and size of data, which contributes to high recognition accuracy. Finally, training and recognition processes. The experiments show that the proposed method of marker selection provides the better results in classification than using all markers.

This study investigates how technical advances in LLMs translate into measurable educational benefit. University admission counseling plays a crucial role in helping prospective students make their higher education decisions. However, traditional advisory methods are constrained by issues such as limited scalability, personalization, and the ability to handle large volumes of inquiries. With the growing need for real-time assistance, Artificial Intelligence (AI), particularly Large Language Models (LLMs), presents a promising solution to these challenges. This paper introduces an AI-driven university admission counseling system that automates routine inquiries, personalizes guidance, and improves accessibility. We develop a formal mathematical framework to represent the counseling task, using embedded and similarity metrics to assess the compatibility of student profiles with academic programs. The system incorporates a multistage workflow for efficient data processing, embedded generation, and AI-driven recommendation. We evaluated the performance of several LLMs eLLAMA, eGPT, and eDEEPSEEK through retrieval-augmented generation (RAG), measuring output quality with NLP metrics such as BLEU, ROUGE, METEOR, and BERTScore. Our results demonstrate that LLMs can significantly improve the efficiency and quality of admission counseling, providing a scalable and adaptable solution that demonstrably enhances student confidence and decision quality.