Noise-to-state practical stability and stabilization of random neural networks