Gas-mixture IR absorption spectra denoising using deep learning

•The MLP DNN filter was developed to decrease noise impact in absorption spectra in the IR range. The results of the MLP DNN filter application were compared with Gaussian filter.•The results show that the MLP DNN reduces the noise essentially more efficiently compared to Gaussian filter. Simultaneously, the MLP DNN filter proved quite accurate restoring the initial pure spectrum shape.•Simulations were conducted for a very high level of noise, which is not typical for the majority of experiments in this field. It allows finding the pessimistic estimations of the MLP DNN filtration accuracy because less level of noise only improves it. It also allows us to emphasize superiority of application of DNNs in preliminary experimental spectral data processing.