Infrared (IR) spectroscopy is a fundamental technique in analytical chemistry. Recently, deep learning (DL) has drawn great interest as the modeling method of infrared spectral data. However, unlike vision or language tasks, IR spectral data modeling is faced with the problem of calibration transfer and has distinctive characteristics. Introducing the prior knowledge of IR spectroscopy could guide the DL methods to learn representations aligned with the domain-invariant characteristics of spectra, and thus improve the performance. Despite such potential, there is a notable absence of DL methods that incorporate such inductive bias. To this end, we propose Analytical-Chemistry-Informed Transformer (ACT) with two modules informed by the field knowledge in analytical chemistry. First, ACT includes learnable spectral processing inspired by chemometrics, which comprises spectral pre-processing, tokenization, and post-processing. Second, a straightforward yet effective representation learning mechanism, namely spectral-attention, is incorporated into ACT. Spectral-attention utilizes the intra-spectral and inter-spectral correlations to extract spectral representations. Empirical results show that ACT has achieved competitive results in 9 analytical tasks covering applications across pharmacy, chemistry, and agriculture. Compared with existing networks, ACT reduces the root mean square error of prediction (RMSEP) by more than 20% in calibration transfer tasks. These results indicate that DL methods in IR spectroscopy could benefit from the integration of prior knowledge in analytical chemistry.

Hyperspectral images provide plentiful spectral–spatial information regarding the nature of different materials, leading to the potential of more efficient detection in diverse areas. However, the high volume of spectral bands, along with limited reference data, leads to many challenges. To alleviate these issues, we propose a superpixel-correlation-based multiview classification approach. Here, the spectral–spatial multiple views are generated via multiscale superpixel segmentation and correlation-based spectral band clustering. A random forest classifier in conjunction with Markov random field regularization is used as the backend classifier of each view. In particular, an energy function with improved metrics of smoothness is introduced. For decision fusion, the pixelwise weight maps of the views are generated based on both classification certainty and neighboring smoothness. The proposed approach is evaluated on three widely used hyperspectral data sets, and the experimental results demonstrate that the proposed method can achieve a competitive performance compared with other existing methods.