Supplemental: Training Neural Networks is NP-Hard in Fixed Dimension A Detailed Proof of NP-Hardness for Two Dimensions-axis (with x 1 = 0, we call this vertical line h