In this talk, we propose a nonlinear multiclass classification-based reduced order model (MC-ROM), for solving time-dependent parametric partial differential equations (PPDEs). This work is inspired by the observation of applying the deep learning-based reduced order model (DL-ROM) to solve diffusion-dominant PPDEs. We find that the generalization ability for one single DL-ROM is not sufficient for diffusion-dominant problem. To improve the generalization ability, we introduce a simple multiclass classification model, for example, SVM, to capture the feature in parameter space (with time) which could be used the classify the input parameters and then apply different DL-ROM models for different cases. When subsets have the same architecture, we can use transfer learning techniques to accelerate offline training. Numerical experiments show that the MC-ROM has a bettern generalization ability than the DL-ROM both for diffusion- and convection-dominant problems, and maintains a good approximation ability. We also compare approximation accuracy and computational efficiency of the proper orthogonal decomposition (POD) which is not suitable for convection-dominant problems. For diffusion-dominant problems, the MC-ROM has better approximation accuracy than the POD in a small dimensionality reduction space, and itscomputational performance is more efficient than the POD's.