Controlling False Discovery in Arbitrarily Structured Hypothesis Spaces via Reproducing Kernels

In three linesNew method to control False Discovery Rate (FDR) in structured hypothesis spaces using reproducing kernels. Reformulates the problem as regularized learning in RKHS, unifying continuous domains, graphs, and hierarchies. Validated on spatial data and differential gene expression tasks.

Summary generated by Claude — human-verified