New Visualization

New visualization techniques are emerging to address the challenges of interpreting complex data and models, particularly in machine learning and AI. Current research focuses on developing interactive and explainable visualizations, including methods for depicting concept drift in time-series data, visualizing AI-discovered patterns in street-level imagery, and decomposing images to explain neural network classifications. These advancements aim to improve understanding of model behavior, facilitate data exploration, and ultimately enhance the usability and trustworthiness of AI systems across diverse scientific and practical applications.