Explainable Metric

Explainable metrics aim to make the decision-making processes of machine learning models more transparent and understandable, addressing the "black box" problem prevalent in many advanced algorithms. Current research focuses on developing such metrics across diverse applications, including medical diagnosis (e.g., using handwriting analysis for neurodegenerative disease detection), model evaluation (e.g., assessing large language model alignment and image synthesis quality), and data bias mitigation. This work leverages various techniques, from handcrafted features and hierarchical semantic segmentation to large language models acting as evaluators, ultimately improving model interpretability and facilitating more reliable and trustworthy AI systems.