The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) Constraint Active Search for Human-in-the-Loop Optimization with Gustavo Malkomes - #505
Jul 29, 2021
Gustavo Malkomes, a research engineer at Intel with expertise in active learning and multi-objective optimization, dives into an innovative algorithm for multiobjective experimental design. He discusses how his work allows teams to explore multiple metrics simultaneously and efficiently, enhancing human-in-the-loop optimization. The conversation covers the balance between competing goals, the significance of stable solutions, and the fascinating applications of his research in real-world scenarios, such as optimization and drug discovery.
AI Snips
Chapters
Books
Transcript
Episode notes
Material Design Collaboration
- Gustavo Malkomes' colleagues collaborated with the University of Pittsburgh to create new glass types for solar panels.
- They used numerical simulations to understand trade-offs in creating nanostructures on glass to improve solar panel efficiency.
Beyond Pareto Efficiency
- Multi-objective experimental design considers multiple metrics in material design.
- The Pareto Efficient Frontier represents optimal trade-offs, but real-world applications require considering parameter stability.
Stability over Optimality
- Optimizing solely for metrics can lead to unstable designs.
- A theoretically less optimal design might be more practical due to its stability and ease of production.
