Quantum computing in general and quantum machine learning in particular provide promising approaches to solve complex problems. Quantum superposition, which is one of the most important characteristics in quantum computing, allows us to embed and process different information sources (e.g. multi-modality or multi-view inputs) simultaneously. In this study, we investigate for the first time multi-feature machine learning based on quantum superposition. We show that when training with superposed quantum data, the quantum models can easily extend their learning space to gain a significant improvement over the single-feature learning.