From Lab to Reality: A Practical Evaluation of Deep Learning Models and LLMs for Vulnerability Detection

A recent study highlights the importance of safety alignment in large language models (LLMs) as they are increasingly adapted for various tasks. The research identifies safety degradation during fine-tuning, attributing it to catastrophic forgetting, and proposes continual learning (CL) strategies to preserve safety. The evaluation of these strategies shows that they can effectively reduce attack success rates compared to standard fine-tuning methods.

A new framework called the Spatiotemporal Reasoning Framework (STAR) has been introduced to enhance the capabilities of Multimodal Large Language Models (MLLMs) in Video Question Answering (VideoQA) tasks. This framework aims to improve the models' ability to understand spatial relationships and temporal dynamics in videos by strategically scheduling tool invocation sequences, thereby enhancing reasoning capabilities.

Recent advancements in KL-Regularized Policy Gradient algorithms have been proposed to enhance the reasoning capabilities of large language models (LLMs). The study introduces a unified derivation known as the Regularized Policy Gradient (RPG) view, which clarifies the necessary weighting for KL variants in off-policy settings, aiming to optimize the surrogate for the intended KL-regularized objective.

A new study introduces the DeepNews Framework, which aims to overcome the limitations of large language models (LLMs) in long-form text generation by addressing the 'Statistical Smoothing Trap.' This framework incorporates cognitive processes similar to those of expert financial journalists, enhancing the quality of generated content.

A new study has introduced a comprehensive pipeline for detecting and mitigating biases in textual data used to train large language models (LLMs), addressing representation bias and stereotypes as mandated by regulations like the European AI Act. The proposed pipeline includes generating word lists, quantifying representation bias, and employing sociolinguistic filtering to mitigate stereotypes.

A recent study highlights the need to address anthropocentric biases in evaluating large language models (LLMs), identifying two overlooked types: auxiliary oversight and mechanistic chauvinism. These biases can hinder the accurate assessment of LLM cognitive capacities, necessitating a more nuanced evaluation approach.

Recent advancements in large language models (LLMs) have led to the development of LLM4FS, a hybrid strategy that combines LLMs with traditional data-driven methods for automated feature selection. This approach evaluates state-of-the-art models like DeepSeek-R1 and GPT-4.5, demonstrating superior performance in selecting relevant features for decision-making tasks.

A novel Deep Learning pipeline has been introduced for establishing feature correspondences in endoscopic image pairs, addressing the challenges of accurate spatial understanding in minimally invasive surgical procedures. This approach focuses on self-supervised contrastive embedding adaptation to enhance image matching capabilities in complex anatomical environments.