A research paper entitled “Achieving joint calibration of soil Vis-NIR spectra across instruments, soil types and properties by an attention-based spectra encoding-spectra/property decoding architecture” by PhD Student Lei Tong and his supervisor Professor Da-Wen Sun, which was published in Geoderma (Volume 405, 1 January 2022, 115449), has been selected as “Editors’ Choice” article for October 2021. Geoderma is a top journal in soil science, which ranks 3rd position among the 37 journals in the soil science category in the Web of Science SCIE edition, with a 2020 impact factor of 6.115.

The recommendation of the Editor-in-Chief, Professor Budiman Minasny, reads: NIR spectroscopy application for soil characterisation has been an active research topic for more than a decade now. Soil spectra libraries from different regions have been compiled. One immediate problem is that different spectrometers produce different soil NIR spectral responses. This issue, while acknowledged, had not been well researched. Tong Lei and Da-Wen Sun from Agriculture and Food Science Centre, University College Dublin utilised a spectra encoding and decoding algorithm (an “attention-based model”) to solve this problem. They demonstrated that they can combine three spectra databases, each collected by a different instrument, to develop a joint calibration model.

The authors state that the major problem for soil Vis-NIR analysis is now solved: “training soil spectral data across different instruments, soil types and properties in one model to achieve joint calibration and analysis”. I look forward to a future application, "federated learning to soil Vis-NIR spectra research to implement an online joint calibration.”

A figure from the article will also form the banner of the journal Twitter account for the month of November 2021.

Lei Tong joined UCD in September 2018 as a PhD student under the supervision of Professor Da-Wen Sun in UCD School of Biosystems and Food Engineering with funding support from the UCD-CSC Scholarship Scheme. In his 4^th year of PhD study, Lei Tong has already published six papers in peer-reviewed international journals:

1. T. Lei and D.-W. Sun, Achieving Joint Calibration of Soil Vis-NIR Spectra Across Instruments, Soil Types and Properties by an Attention-Based Spectra Encoding-Spectra/Property Decoding Architecture, Geoderma, 405 (2022) 115449.
2. T. Lei, B. Tobin, Z.-H. Liu, S.-Y. Yang and D.-W. Sun, A Terahertz Time-Domain Super-Resolution Imaging Method Using a Local-Pixel Graph Neural Network for Biological Products, Analytica Chimica Acta, 1181 (2021) 338898.
3. T. Lei and D.-W. Sun, A Novel NIR Spectral Calibration Method: Sparse Coefficients Wavelength Selection and Regression (SCWR), Analytica Chimica Acta, 1110 (2020) 169-180.
4. T. Lei, X.-H. Lin and D.-W. Sun, Rapid Classification of Commercial Cheddar Cheeses from Different Brands Using PLSDA, LDA and SPA-LDA Models Built by Hyperspectral Data, Journal of Food Measurement and Characterization, 13 [4] (2019) 3119-3129.
5. T. Lei and D.-W. Sun, Developments of Nondestructive Techniques for Evaluating Quality Attributes of Cheeses: A Review, Trends in Food Science & Technology, 88 (2019) 527-542.
6. Y. Ren, X. Lin, T. Lei and D.-W. Sun, Recent Developments in Vibrational Spectral Analyses for Dynamically Assessing and Monitoring Food Dehydration Processes, Critical Reviews in Food Science and Nutrition, (2022) DOI: https://doi.org/10.1080/10408398.2021.1947773.

The paper (opens in a new window)Achieving joint calibration of soil Vis-NIR spectra across instruments, soil types and properties by an attention-based spectra encoding-spectra/property decoding architecture, Geoderma, Volume 405, 1 January 2022, 115449 can be downloaded free of charge by clicking (opens in a new window)here. You can access “October 2021 Editors' Choice” of Geoderma by clicking (opens in a new window)here.