Recently, with the advancement of technology, the automotive industry has seen an increase in network connectivity. CAN (Controller Area Network) bus technology enables fast and efficient data communication between various electronic devices and systems within a vehicle, providing a platform that integrates and manages a wide range of functions, from core systems to auxiliary features. However, this increased connectivity raises concerns about network security, as external attackers could potentially gain access to the automotive network, taking control of the vehicle or stealing personal information. This paper analyzed abnormal messages occurring in CAN and confirmed that message occurrence periodicity, frequency, and data changes are important factors in the detection of abnormal messages. Through DBC decoding, the specific meanings of CAN messages were interpreted. Based on this, a model for classifying abnormalities was proposed using the GRU model to analyze the periodicity and trend of message occurrences by measuring the difference (residual) between the predicted and actual messages occurring within a certain period as an abnormality metric. Additionally, for multi-class classification of attack techniques on abnormal messages, a Random Forest model was introduced as a multi-classifier using message occurrence frequency, periodicity, and residuals, achieving improved performance. This model achieved a high accuracy of over 99% in detecting abnormal messages and demonstrated superior performance compared to other existing models.

The adoption of cloud environments has revolutionized application deployment and management, with microservices architecture and container technology serving as key enablers of this transformation. However, these advancements have introduced new challenges, particularly the necessity to precisely understand service interactions and conduct detailed analyses of internal processes within complex service environments such as microservices. Traditional monitoring techniques have proven inadequate in effectively analyzing these complex environments, leading to increased interest in eBPF (extended Berkeley Packet Filter) technology as a solution. eBPF is a powerful tool capable of real-time event collection and analysis within the Linux kernel, enabling the monitoring of various events, including file system activities within the kernel space. This paper proposes a container activity analysis system based on eBPF, which monitors events occurring in the kernel space of both containers and host systems in real-time and analyzes the collected data. Furthermore, this paper conducts a comparative analysis of prominent eBPF-based container monitoring systems (Tetragon, Falco, and Tracee), focusing on aspects such as event detection methods, default policy application, event type identification, and system call blocking and alert generation. Through this evaluation, the paper identifies the strengths and weaknesses of each system and determines the necessary features for effective container process monitoring and restriction. In addition, the proposed system is evaluated in terms of container metadata collection, internal activity monitoring, and system metadata integration, and the effectiveness and future potential of eBPF-based monitoring systems.

RISC-V is an open-source instruction set architecture, used in various hardware implementations, and can be flexibly expanded to meet system requirements through the RV64I base instruction set and 16 standard extensions. Currently, the RISC-V architecture employs the shadow stack technique to protect return addresses. This paper compares the performance of the compact shadow stack mechanism and the parallel shadow stack mechanism in the RISC-V architecture using the SPEC CPU 2017 and beebs benchmarks. Experimental results show that the parallel shadow stack mechanism exhibits higher overhead than the compact shadow stack mechanism. This suggests that the efficiency of the parallel mechanism is reduced due to the limitations of the RISC-V architecture, making the compact shadow stack more suitable for RISC-V. Additionally, this paper identifies the security limitations of the existing RISC-V shadow stack and proposes directions for enhancing the performance and security of shadow stack mechanisms to ensure a secure execution environment for RISC-V.

Recently, services utilizing large-scale language models (LLMs) such as GPT-4 and LLaMA have been released, garnering significant attention. These models can respond fluently to various user queries, but their insufficient training on Korean data raises concerns about the potential to provide inaccurate information regarding Korean culture and language. In this study, we selected eight major publicly available models that have been trained on Korean data and evaluated their understanding of Korean culture using a dataset composed of five domains (Korean language comprehension and cultural aspects). The results showed that the commercial model HyperClovaX exhibited the best performance across all domains. Among the publicly available models, Bookworm demonstrated superior Korean language proficiency. Additionally, the LDCC-SOLAR model excelled in areas related to understanding Korean culture and language.

As Large Language Models (LLMs) continue to advance, effectively harnessing their potential has become increasingly important. LLMs, trained on vast datasets, are capable of generating text across a wide range of topics, making them useful in applications such as content creation, machine translation, and chatbots. However, they often face challenges in generalization due to gaps in specific or specialized knowledge, and updating these models with the latest information post-training remains a significant hurdle. To address these issues, Retrieval-Augmented Generation (RAG) models have been introduced. These models enhance response generation by retrieving information from continuously updated external databases, thereby reducing the hallucination phenomenon often seen in LLMs while improving efficiency and accuracy. This paper presents the foundational architecture of RAG, reviews recent research trends aimed at enhancing the retrieval capabilities of LLMs through RAG, and discusses evaluation techniques. Additionally, it explores performance optimization and real-world applications of RAG in various industries. Through this analysis, the paper aims to propose future research directions for the continued development of RAG models.

Adhering to coding style guidelines is crucial for both companies and developers as it improves code readability and reduces the costs associated with testing and maintenance. However, teaching coding style in programming courses poses challenges. Setting up an environment for learning coding styles is hard, and there are no predefined coding style rules for beginners. From the learners' perspective, since adherence to coding styles does not affect their grades, they do not feel a strong need to learn them. This paper introduces a coding style checking system for an online evaluation system. The proposed system is implemented to check and evaluate coding styles in C, Java, and Python. Additionally, we applied 234 out of the 1,023 rules provided by the language-specific tools, which is 23.08%, allowing for the application of coding style rules according to the course progression. Moreover, we motivated learners to improve their coding style by adding quality scores to their basic scores. After introducing the coding style education system, the number of students scoring over 25 points on their initial submissions increased by 149.47%, from 18 students in the first week to 44 students in the sixth week. Learners used the coding style checking system to learn how to apply coding style rules and subsequently implemented their code in adherence to the specified coding styles.

This study proposes a deep learning architecture optimized for fire detection derived through Layer Importance Evaluation. In order to solve the problem of unnecessary complexity and operation of the existing Convolutional Neural Network (CNN)-based fire detection system, the operation of the inner layer of the model based on the weight and activation values was analyzed through the Layer Importance Evaluation technique, the layer with a high contribution to fire detection was identified, and the model was reconstructed only with the identified layer, and the performance indicators were compared and analyzed with the existing model. After learning the fire data using four transfer learning models: Xception, VGG19, ResNet, and EfficientNetB5, the Layer Importance Evaluation technique was applied to analyze the weight and activation value of each layer, and then a new model was constructed by selecting the top rank layers with the highest contribution. As a result of the study, it was confirmed that the implemented architecture maintains the same performance with parameters that are about 80% lighter than the existing model, and can contribute to increasing the efficiency of fire monitoring equipment by outputting the same performance in accuracy, loss, and confusion matrix indicators compared to conventional complex transfer learning models while having a learning speed of about 3 to 5 times faster.

Recent advances in generative AI technologies using natural language processing have critically impacted text, image, and video production. Despite these innovations, we still need to improve the consistency and reusability of AI-generated outputs. These issues are critical in cartoon creation, where the inability to consistently replicate characters and specific objects can degrade the work's quality. We propose an integrated adaption of language analysis-based requirement engineering and cartoon engineering to solve this. The proposed method applies the linguistic frameworks of Chomsky and Fillmore to analyze natural language and utilizes UML sequence models for generating consistent 3D representations of object interactions. It systematically interprets the creator's intentions from textual inputs, ensuring that each character or object, once conceptualized, is accurately replicated across various panels and episodes to preserve visual and contextual integrity. This technique enhances the accuracy and consistency of character portrayals in animated contexts, aligning closely with the initial specifications. Consequently, this method holds potential applicability in other domains requiring the translation of complex textual descriptions into visual representations.

Targeted career development is critical for student success but is often lacking for underrepresented students at many public higher-education institutions due to insufficient career counseling resources. We propose an innovative career development tool leveraging Large Language Models (LLMs) to enhance student career prospects through three steps: (1) identifying relevant jobs by analyzing resumes, (2) pinpointing skill gaps using external resources such as classroom assignments, in addition to resumes, and (3) suggesting customized learning paths. Our tool accurately matches jobs in real-world settings, identifies true skill gaps while reducing false positives, and provides learning paths that receive high satisfaction scores from faculty. Future research will enhance the solution's capabilities by incorporating diverse external resources and leveraging advancements in LLM technology to better support early-stage career seekers.

Excavation work in urban areas can induce ground deformation, which may damage nearby infrastructure. Such ground deformation can result in displacement of paving blocks near the construction site. Accurate measurement of these displacements can serve as an indicator for assessing the potential risks associated with ground deformation. This paper proposes a robust and efficient method for paving block displacement measurement using a 3D laser line sensor mounted on an Unmanned Ground Vehicle (UGV). The proposed method consists of two stages: 2D projection based object detection and measurement through the CPLF algorithm. Experimental results demonstrate that the CPLF algorithm is more efficient compared to the PLF algorithm, achieving an error of 1.36 mm and a processing time of 10.76 ms, confirming that the proposed method ensures robust online measurements with high accuracy in real-world environments with various types of paving blocks and environmental factors using a 3D laser line sensor on a UGV.