In recent years, large language models (LLMs) have gained significant traction across various domains, including education. This paper explores the application of LLMs in grading programming assignments. By leveraging data collected from existing programming assignments and their corresponding grades, we aim to develop a robust LLM-based grading system. We also incorporate augmented data representing various grading scenarios to enhance the model’s performance and ensure comprehensive coverage across all grading levels. Our approach involves training the LLM on this combined dataset to enable accurate and consistent evaluation of programming assignments. The proposed model, BeGrading, aims to reduce the grading burden on educators and provide timely and objective feedback to students. Compared to the Codestral model, our proposed model demonstrates an absolute difference rate of 19%, equivalent to $$\pm 0.95$$ ± 0.95 out of 5. This is acceptable for using a small, fine-tuned model with optimized data. Additionally, the Codestral model compared to the dataset optimized score shows a difference of 15% equivalent to a margin of $$\pm 0.75$$ ± 0.75 out of 5. Preliminary results demonstrate the potential of LLMs to perform grading tasks with a high degree of reliability, opening avenues for further research and practical applications in automated education systems.

The growing demand for radiological services, amplified by a shortage of qualified radiologists, has resulted in significant challenges in managing the increasing workload while ensuring the accuracy and timeliness of radiological reports. To address these issues, recent advancements in artificial intelligence (AI), particularly in transformer models, vision-language models (VLMs), and Large Language Models (LLMs), have emerged as promising solutions for radiology report generation (RRG). These systems aim to make diagnosis faster, reduce the workload for radiologists by handling routine tasks, and help generate high-quality, consistent reports that support better clinical decision-making. This comprehensive study covers RRG developments from 2021 to 2025, focusing on emerging transformer-based and VLMs, highlighting the key methods, architectures, and techniques employed. We examine the datasets currently available for RRG applications and the evaluation metrics commonly used to assess model performance. In addition, the study analyzes the performance of the leading models in the field, identifying the top performers and offering insights into their strengths and limitations. Finally, this study proposes new directions for future research, emphasizing potential improvements to existing systems and exploring new avenues for advancing the capabilities of AI in radiology report generation.

Knowledge Graphs (KGs) are becoming an essential tool in the organization, linkage, and analysis of complex data across various domains. By structuring information as entities and their interrelationships, KGs enhance data integration and provide advanced capabilities such as reasoning, question-answering, and real-time decision-making. This paper explores methodologies for building and maintaining static and dynamic knowledge graphs, focusing on their applications in fields like virtual assistants, recommendation systems, autonomous systems, and AI-driven decision-making. The study emphasizes the growing significance of automation tools such as Named Entity Recognition (NER) and relationship extraction and addresses challenges in balancing automation with accuracy in KG construction. Furthermore, the transformation of static knowledge graphs into dynamic, continuously updating systems is discussed, showcasing their potential to revolutionize real-world applications by integrating ever-evolving datasets. In this context, dynamic KGs excel in domains that require real-time updates, making them invaluable for climate monitoring, autonomous vehicles, and production fault detection systems. This study provides insights into the methodologies and tools that can ensure high-quality KG creation, the challenges of real-time data integration, and how KGs contribute to advancing AI capabilities in data-rich environments.

Conventional Linux system call filters validate op-erations statelessly, leaving applications vulnerable to sequence-based exploits. While sequence-aware modeling is common in in-trusion detection, existing solutions typically offer passive alertingor require manual policy generation. We present a frameworkbridging behavioral learning and runtime enforcement that islightweight in terms of its compact behavioral representationand bounded lookup complexity. Our decoupled architectureleverages eBPF to automatically extract system call and processlifecycle transitions under benign workloads, synthesizing adeterministic first-order finite-state machine (FSM). This baselineis compiled into a structurally validated binary policy artifact andloaded into the kernel. At runtime, the eBPF engine validatesexecution sequences with constant O(1) algorithmic complexity,employing a dual-hook strategy: synchronously blocking sensitiveoperations via LSM hooks, and asynchronously terminatingthread groups via raw tracepoints shortly after other violationscommence. Evaluation on synthetic and production server work-loads yielded no false-positive violations across evaluated benignruns, 100% detection across the five evaluated repetitions ofthe tested threat scenario, and median end-to-end detection-onlysyscall latencies below 25 μs for evaluated filesystem operationson the testbed, demonstrating the feasibility of sequence-awareapplication isolation under scoped behaviors