1087-ProtoCloud: Explainable Model for Single-Cell Analysis
ProtoCloud is a sophisticated deep learning model designed to enhance the accuracy and transparency of single-cell genomic analysis. By organizing data around cell-type-specific prototypes, the system provides dual-level interpretability that explains both individual cell classifications and the specific genes driving those decisions. A unique disentangled latent space allows the model to separate essential biological identities from technical noise, such as batch effects, without requiring external metadata. Beyond standard annotation, the model features an uncertainty estimation mechanism…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.