Abstract
In spectral data predictive modelling of fresh fruit, often the models are calibrated to predict multiple responses. A common method to deal with such a multi-response predictive modelling is the partial least-squares (PLS2) regression. Recently, deep learning (DL) has shown to outperform partial least-squares (PLS) approaches for single fruit traits prediction. The DL can also be adapted to perform multi-response modelling. This study presents an implementation of DL modelling for multi-response prediction for spectral data of fresh fruit. To show this, a real NIR data set related to SSC and MC measurements in pear fruit was used. Since DL models perform better with larger data sets, a data augmentation procedure was performed prior to data modelling. Furthermore, a comparative study was also performed between two of the most used DL architectures for spectral analysis, their multi-output and single-output variants and a classic baseline model using PLS2. A key point to note that all the DL modelling presented in this study is performed using novel automated optimisation tools such as Bayesian optimisation and Hyperband. The results showed that DL models can be easily adapted by changing the output of the fully connected layers to perform multi-response modelling. In comparison to the PLS2, the multi-response DL model showed ∼13 % lower root mean squared error (RMSE), showing the ease and superiority of handling multi-response by DL models for spectral calibration.
Highlights
Multivariate data acquired with advanced analytical instruments such as spectrometers are rich in chemical and physical information related to the fruit analysed (Mishra et al, 2020; Saeys et al, 2019)
PLS2 optimization was done by fitting multiple partial least-squares (PLS) models with varying number of Latent Variables (LVs) in the range (1–50) to the calibration set and computing their root mean squared error (RMSE) on the calibration and tuning sets
This study demonstrated the potential of the deep learning (DL) models to multiresponse modelling for fruit traits prediction by changing the final output layer of the DL to the desired number of responses
Summary
Multivariate data acquired with advanced analytical instruments such as spectrometers are rich in chemical and physical information related to the fruit analysed (Mishra et al, 2020; Saeys et al, 2019). The visible and near-infrared (Vis-NIR) spectroscopy data consists of colour information, overtones of fundamental chemical bonds such as OH, CH, NH and SH, and light scattering information related to the physical structure of the fruit (Nicolai et al, 2007) Due to such a rich information captured by advanced analytical instruments, often, the user is interested to reap the full benefit of the information and try to predict multiple chemicals as well as physical properties in the samples (Mishra et al, 2021b). In the case of fresh fruit analysis with diffuse reflectance near-infrared (NIR) spectroscopy, NIR calibrations are widely used to predict several chemical properties such as moisture content (MC) and soluble solids content (SSC) (Walsh et al, 2020) Another example of where multi-output NIR model was found to be helpful was related to the prediction of several types of fats and proteins in meat produce (Zomeno et al, 2012) and several soil prop erties (Ng et al, 2019). The need to predict multiple response variables with a single model explains the need to develop multi-response predictor models
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
