Nowadays, machine olfaction has been widely used in many fields. In this paper, we proposed an odor labeling convolutional encoder-decoder (OLCE) for odor sensing. In order to demonstrate the performance of the model, a comparison of several algorithms that had been applied to machine olfaction was implemented. Experiment results demonstrated that OLCE had a decent performance using in machine olfaction. The challenge of establishing general odor characterization was discussed.