nx_pose_elm.nx

nx_pose_elm.nx -- the NONLINEAR learner for pose-from-silhouette (roadmap R2 v1). The linear perceptron on raw occupancy provably can't separate the hand regions (tried 9-region, delta, 3-band = all at chance). An EXTREME LEARNING MACHINE fixes that WITHOUT fragile fixed-point backprop: a FIXED RANDOM projection h = relu(R*x + b) lifts the occupancy into a high-dim nonlinear feature space where the classes ARE linearly separable, and only the LINEAR OUTPUT is trained -- by the PROVEN perceptron rule (same one text2motion used). Random-features / ELM is a real ML method (Rahimi-Recht random kitchen sinks; Huang ELM). Seeded LCG => deterministic. The honest ceiling stays: coarse, our-render-silhouette domain; SOTA pose = a trained CNN. license_tier: ORIGINAL