Differential network analysis is an important statistical problem with wide applications. Many statisticians focus on binary problems and propose to perform such analysis by obtaining sparse estimates of the difference between precision matrices. These methods are supported by excellent theoretical properties and practical performance. However, efficient computation for these methods remains a challenging problem. A novel algorithm referred to as the SMORE algorithm is proposed for differential network analysis. The SMORE algorithm has low storage cost and high computation speed, especially in the presence of strong sparsity. In the meantime, the SMORE algorithm provides a unified framework for binary and multiple network problems. In addition, the SMORE algorithm can be applied in high-dimensional quadratic discriminant analysis problems as well, leading to a new approach for multiclass high-dimensional quadratic discriminant analysis. Numerical studies confirm the stability and the efficiency of the proposed SMORE algorithm in both differential network analysis and quadratic discriminant analysis.