From Stochastic Conjugate Gradient to Stochastic Second-Order Optimization, Driven by Conjugate Coefficient With Second-Order Information | Synapse