Eugen Harinda

This review paper provides a comprehensive analysis of recent advancements in leveraging reinforcement learning (RL) and deep reinforcement learning (DRL) techniques for optimizing traffic signal control systems within urban environments. Traditional traffic signal management methods often fail to address the inefficiencies resulting in congestion, delays, environmental impacts, and driver dissatisfaction. In contrast, RL methodologies empower traffic signals to dynamically learn and refine signal timing strategies by iteratively interacting with real-world traffic conditions. DRL builds upon RL by integrating deep neural networks to better capture and respond to intricate traffic patterns. The paper investigates the potential of RL and DRL approaches in addressing diverse urban traffic challenges across peak and non-peak periods, as well as in areas with irregular traffic distributions. It delves into the strengths, weaknesses, underlying principles, and advanced architectures of current RL/DRL methodologies proposed for this domain. Furthermore, it evaluates practical applications and simulation studies, showcasing performance enhancements over traditional signal control methods. In conclusion, the review consolidates key findings and outlines future research directions for implementing RL/DRL systems in real-world scenarios to effectively alleviate urban traffic congestion.

As Large Language Models (LLMs) continue to advance in their ability to process natural language, their potential to transform industries and reshape the future of human-computer interaction becomes increasingly evident. This study evaluates the application of Large Language Models (LLM) to automate SQL query generation from natural language inputs in enterprise environments. We investigated the feasibility of using open-source LLMs, including Mistral, CodeLlama, Phi-3, and DeepSeek Coder, by fine-tuning them with a custom dataset reflecting company-specific data tables. This dataset was iteratively constructed using a baseline LLM that allows fine-tuning to address unique enterprise data structures and use cases. A two-stage filtering and refinement mechanism was implemented to improve query accuracy. The first stage identifies relevant tables and the second stage adds an iterative error correction step to improve the SQL query generation process. The resulting system significantly reduced query errors and increased accuracy by 84%,88%,81%, and 90% for Mistral, CodeLlama, Phi-3, and DeepSeek Coder LLM. However, challenges remain in resource consumption and logical error handling. Future work will focus on improving contextual understanding and integrating advanced AI techniques to further enhance robustness and applicability.