3 papers
cs.LG2026
Blind Thermodynamic Ontology Discovery from Anonymous Experiments
Linzhe Zhang, Changming Xu
Before a machine learning model can learn a thermodynamic equation of state, it must discover what its measurements represent: which channels scale with system size, which are inte…
cs.LG2026
Discovering Physical Representation Languages
Linzhe Zhang, Changming Xu
Before a machine can discover a physical law, it must discover what its measurements are: which observations live on cells, which are intensive or extensive, which sectors are dual…
cs.LG2026
What Can a Recurrent State Safely Forget?
Linzhe Zhang, Changming Xu
Recurrent models must preserve information that changes future behavior while suppressing hidden-state error. These objectives conflict: contraction improves stability, but contrac…