Machine learning is rapidly transforming cryospheric science, creating new opportunities to observe change, infer physical processes, emulate complex models, and improve prediction across the Earth system. This session invites contributions that develop or apply machine learning across the cryosphere, including sea ice, land ice, snow, permafrost, ice sheets, ice shelves and other related components. We welcome a broad range of applications, including - but not limited to – remote sensing, emulation of physical models, physics-informed machine learning, causal inference, calibration and initialization, uncertainty quantification, automated process discovery, and the development  of novel datasets and benchmarks. Relevant approaches include classical statistical modelling, Bayesian inference frameworks including amortized inference and generative approaches, deep learning, and other data-driven frameworks.  We particularly encourage studies that couple methodological innovation with new scientific insight, improved prediction, or stronger links between observations, process understanding, and modeling across the cryosphere.