Beyond Numerical Features: CNN-Driven Algorithm Selection via Contour Plots for Continuous Black-Box Optimization
arXiv:2605.20797v1 Announce Type: new Abstract: The present paper introduces a new representation-driven approach to per-instance algorithm selection, applied to black-box optimization, for automatically choosing the most promising solver from a fixed portfolio. Prior work in continuous optimization largely relies on numerical descriptors, including Exploratory Landscape Analysis features and learned embeddings such as Deep-ELA. This work studies a complementary representation: contour-map visualizations of probed landscapes. A CNN regressor takes multiple instance-specific contour views (stac
The continuous drive for more efficient and automated AI model selection, coupled with advancements in computer vision and graph-based representations of data, enables novel approaches like CNN-driven technique selection.
This development enhances the efficiency and autonomy of AI systems by improving automatic algorithm selection for complex optimization problems, reducing manual intervention and increasing performance.
Algorithm selection for black-box optimization can now leverage visual representations like contour plots, potentially leading to more accurate and robust solver choices than purely numerical or embedded feature approaches.
- · AI/ML researchers
- · Developers of AutoML platforms
- · Industries using black-box optimization
- · Generative AI
- · Manual algorithm selection processes
Improved performance and broader applicability of automated machine learning solutions will be observed.
This methodology could be generalized to other areas of AI where visual data representations offer an advantage for problem-solving or system configuration.
Increased autonomy in complex system design could further accelerate the development of advanced AI agents by reducing bottlenecks in their problem-solving capabilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG