What question did this study set out to answer?

This research aims to develop a rapid method using near-infrared spectroscopy to detect key indicators and verify the authenticity of beer.

March 28, 2026MoleculesOpen Access

Key Indicator Detection and Authenticity Identification of Beer Based on Near-Infrared Spectroscopy Combined with Multi-Task Feature Extraction

View Full Paper

Ask AI

Bookmark

Authors

Proposes a rapid NIRS method for detecting key indicators and verifying authenticity in beer, suggesting significant regulatory applications.

This research aims to develop a rapid method using near-infrared spectroscopy to detect key indicators and verify the authenticity of beer.
Utilized variable importance in projection for wavelength selection.
Employed multi-task learning strategies along with convolutional neural networks and long short-term memory networks.
Optimized model hyperparameters using a Bayesian optimization algorithm.
Established partial least squares regression and support vector machine regression models.
The MTL-based CNN-LSTM-MHA network significantly improved model generalization.
Achieved coefficients of determination (R2) of 0.996 and 0.997 for alcohol content and original wort concentration, respectively.
In an independent test set, R2 values were 0.995 and 0.991 with relative root mean square errors of 2.515% and 2.087%.
Achieved 100% classification accuracy across all datasets.

Wei et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

March 28, 2026MoleculesOpen Access

View Full Paper

Ask AI

Bookmark

Authors

Proposes a rapid NIRS method for detecting key indicators and verifying authenticity in beer, suggesting significant regulatory applications.

This research aims to develop a rapid method using near-infrared spectroscopy to detect key indicators and verify the authenticity of beer.
Utilized variable importance in projection for wavelength selection.
Employed multi-task learning strategies along with convolutional neural networks and long short-term memory networks.
Optimized model hyperparameters using a Bayesian optimization algorithm.
Established partial least squares regression and support vector machine regression models.
The MTL-based CNN-LSTM-MHA network significantly improved model generalization.
Achieved coefficients of determination (R2) of 0.996 and 0.997 for alcohol content and original wort concentration, respectively.
In an independent test set, R2 values were 0.995 and 0.991 with relative root mean square errors of 2.515% and 2.087%.
Achieved 100% classification accuracy across all datasets.

Wei et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark