Articles
Technical articles & explainers
Notes from Chun-Ming Huang's research and engineering in seismic AI, explaining the principles behind models, loss functions, and real-world data in plain language.
How Did AI Learn to Pick Seismic Phases?
In just a few years, AI phase picking moved from each lab writing its own code to loading a trained model with a single command. This article introduces three milestones: PhaseNet, EQTransformer, and SeisBench, which unified the ecosystem.
Read more →Why Does AI Often Miss S-Waves?
Deep-learning phase pickers have a long-standing problem: the S-wave position is correct, but its amplitude is pushed below the detection threshold and counted as undetected. This article explains the cause and how BlueDisc recovers these S-waves with a shape-aware loss, increasing effective detections by 64%.
Read more →