Infinite-dimensional operator ridgelet transform

1. Neural networks with Hilbert-space inputs🔗

This chapter is Section 2 of the manuscript together with Appendix F. The input space H is a separable real Hilbert space, the output space Y a separable complex Hilbert space, and a network is a superposition of ridge functions x\mapsto\beta(\langle a,x\rangle+c). Section 2 defines the finite-width and the integral networks; Appendix F shows that networks with operator-valued parameters reduce exactly to them.

  1. 1.1. Finite-width and integral networks
  2. 1.2. Operator-valued parameters