ML4VIS

ML4VIS maps machine learning techniques to visualization processes to characterize and organize research on how ML can enhance data visualization.

Key Features:

• Comprehensive Survey of 88 studies: Systematic survey documenting 88 ML4VIS studies and their methods and applications.
• Structured Understanding: Presents a framework categorizing seven processes—Data Processing4VIS, Data-VIS Mapping, Insight Communication, Style Imitation, VIS Interaction, VIS Reading, and User Profiling—where ML techniques can contribute.
• Integration with Visualization Theories: Aligns those processes with existing visualization theoretical models within an ML4VIS pipeline.
• Alignment with ML Tasks: Maps visualization processes to main learning tasks in machine learning to show correspondences between visualization needs and ML capabilities.
• Current Practices and Future Opportunities: Identifies current practices and potential future directions in ML4VIS research.

Scientific Applications:

• Design and Development: Support visualization design and development by identifying ML approaches relevant to visualization tasks.
• Evaluation: Aid evaluation of visualization tools and methods by relating ML techniques to evaluation objectives.
• User Profiling: Enable analysis of user interactions and preferences through ML for user profiling.

Methodology:

Systematic survey approach to gather and categorize existing research, combined with mapping visualization processes to machine learning tasks.