The feedback nonlinear output-error system is a special nonlinear system, the existence of the memoryless nonlinear block on the feedback channel leads to the difficulty of the parameter estimation. Combining the hierarchical identification principle with the auxiliary model identification idea, we derive an auxiliary model-based hierarchical stochastic gradient algorithm. In order to further improve the convergence rate and parameter estimation accuracy, an auxiliary model-based hierarchical multi-innovation stochastic gradient algorithm is proposed by using the multi-innovation identification theory. Furthermore, the convergence properties of the proposed algorithms are analyzed through the stochastic process theory. Finally, the experimental results indicate the effectiveness of the proposed algorithms.