Eric Mottay from h-nu gave two presentations at Photonics West 2026, both on machine learning for laser micro-processing.
“Machine learning-assisted laser thinning of silicon using integrated multimodal online diagnostics”, presented in collaboration with ALPhANOV, showed that several inexpensive sensors recording during ablation each predict depth accurately enough to guide the process, and that combining them preserves accuracy when one sensor degrades.
“Hybrid analytical-machine learning model for data-efficient prediction of ultrafast laser ablation” described a model combining the two-temperature model with a neural network, which reaches useful accuracy on a few tens of measurements where machine learning alone needs several hundred.
Eric Mottay’s account of the week appeared in the EPIC newsletter.