Backprop lets you calculate the gradient efficiently. What you do with that gradient is up to you (I would try L-BFGS or something akin to a stochastic variatio

Backprop lets you calculate the gradient efficiently. What you do with that gradient is up to you (I would try L-BFGS or something akin to a stochastic variation). So you could use conjugate gradient or some other optimization method