Domain invariant covariate selection (Di-CovSel) for selecting generalized features across domains

Diaz, Valeria Fonseca; Mishra, Puneet; Roger, Jean‐Michel; Saeys, Wouter

doi:10.1016/j.chemolab.2022.104499

Cited by 7 publications

(9 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First, a descriptive analysis of the capacity to align the variability of the domains was done by visualizing the projection scores from all models, including a PCA analysis as a benchmark. After this, the predictive power of all models on the nine target domains was compared in terms of the root mean squared error of prediction (RMSEP),

R_{p}^{2}

, prediction bias (Bias) and the standard error of prediction (SEP) 20 . Notably, we assessed the performance of each model using the same test set split that was used in Anderson et al 22 and Passos and Mishra, 23 that is, including samples from Season 4, to investigate “out‐of‐domain” generalization.…”

Section: Methodsmentioning

confidence: 99%

“…In that case, the differences between the domains are penalized too much and model generalization may suffer. In practice, an optimal trade‐off between explaining a large portion of the variance in

bold-italicy

and domain invariability can be found by tuning

λ

by means of cross‐validation 20 …”

Section: Theorymentioning

confidence: 99%

“…Domain‐adaptive PLS (da‐PLS) is a recent extension that offers the ability to handle more complex dataset shifts by extracting latent variables not only from the original space but also from the reproducing kernel Hilbert space 19 . di‐CovSel combines di‐PLS with covariate selection to identify variables that generalize well across domains 20 …”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Partial least squares regression with multiple domains

Mikulasek

Diaz

Gabauer

et al. 2023

Journal of Chemometrics

View full text Add to dashboard Cite

This paper introduces the multiple domain‐invariant partial least squares (mdi‐PLS) method, which generalizes the recently introduced domain‐invariant partial least squares method (di‐PLS). In contrast to di‐PLS which solely allows transferring of knowledge from a single source to a single target domain, the proposed approach enables the incorporation of data from an arbitrary number of domains. Additionally, mdi‐PLS offers a high level of flexibility by accepting labeled (supervised) and unlabeled (unsupervised) data to cope with dataset shifts. We demonstrate the application of the mdi‐PLS method on a simulated and one real‐world dataset. Our results show a clear outperformance of both PLS and di‐PLS when data from multiple related domains are available for training multivariate calibration models underpinning the benefit of mdi‐PLS.

show abstract

R_{p}^{2}

Section: Methodsmentioning

confidence: 99%

bold-italicy

and domain invariability can be found by tuning

λ

by means of cross‐validation 20 …”

Section: Theorymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Partial least squares regression with multiple domains

Mikulasek

Diaz

Gabauer

et al. 2023

Journal of Chemometrics

View full text Add to dashboard Cite

show abstract

“…For this task, domain adaptation can be used to train an ML model on the spectra acquired under stationary conditions (the source domain), akin to a manufacturer collecting offline data in a laboratory, and aid transfer of these models to the moving conditions on a conveyor line (the task domain). Matrix or kernel-based methods have previously been used for domain adaptation of 1D NIR spectra to find projections between domains [ 21 , 22 ] or to also extract discriminative features to be used for a main learning task [ 23 , 24 , 25 , 26 ]. Ensemble methods that maximize model diversity and prediction similarity in the new domain data have also been used to increase the probability that transferable correlations are used [ 27 ].…”

Section: Introductionmentioning

confidence: 99%

Domain Adaptation for In-Line Allergen Classification of Agri-Food Powders Using Near-Infrared Spectroscopy

Bowler

Ozturk

Rady

et al. 2022

Sensors

View full text Add to dashboard Cite

The addition of incorrect agri-food powders to a production line due to human error is a large safety concern in food and drink manufacturing, owing to incorporation of allergens in the final product. This work combines near-infrared spectroscopy with machine-learning models for early detection of this problem. Specifically, domain adaptation is used to transfer models from spectra acquired under stationary conditions to moving samples, thereby minimizing the volume of labelled data required to collect on a production line. Two deep-learning domain-adaptation methodologies are used: domain-adversarial neural networks and semisupervised generative adversarial neural networks. Overall, accuracy of up to 96.0% was achieved using no labelled data from the target domain moving spectra, and up to 99.68% was achieved when incorporating a single labelled data instance for each material into model training. Using both domain-adaptation methodologies together achieved the highest prediction accuracies on average, as did combining measurements from two near-infrared spectroscopy sensors with different wavelength ranges. Ensemble methods were used to further increase model accuracy and provide quantification of model uncertainty, and a feature-permutation method was used for global interpretability of the models.

show abstract

“…In this context, efforts were devoted to the development of analytical solutions that allow to move the knowledge acquired from one site or data source to another where our interest is focused, such as in the activities of scale-up, product transfer, and process monitoring. , A variety of model transfer approaches were developed for handling specifically spectroscopic data. − Techniques such as partial least-squares (PLS) model inversion, Joint-Y PLS, calibration/model transfer, ,, and, more recently, transfer learning ,,, and domain adaptation offer different paths to achieve the aforementioned goals. These methods address partially the aforementioned challenge of connecting multiple sites, but their objective is not to provide a global management platform but instead a way to transfer existing knowledge to a new context of interest.…”

Section: Introductionmentioning

confidence: 99%

A Federated Classification Approach of Waste Lubricant Oils in Geographically Distributed Laboratories

Reis

Strelet

Sansana

et al. 2022

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Decision making and operational management of multiple identical units, equipment, or plants (sources) that are geographically distributed are becoming increasingly common. Instead of developing a classifier for each source, with much less data and representativeness of the entire population, this work presents a unique centralized classifier using a federated approach. The centralized classifier can take into account all data collected from multiple distributed sources and accommodate their local and specific data structures and correlation patterns. The federated approach has the built-in capability of expansion, allowing for the inclusion of more sources, without the need for retraining. Therefore, new sources may immediately benefit from the existing unified classification model as soon as they are “connected,” which only requires a projection operation. The proposed federated approach is applied to a real case study of predicting an important property (coagulation) of waste lubricant oil (WLO) in several locations of the recovery network. The coagulation behavior determines if WLO can be regenerated and used to produce again base oil. Fourier transform infrared spectra are collected at different sources (laboratories) and combined using the federated framework, leading to high classification accuracy and a more generalizable model.

show abstract

Domain invariant covariate selection (Di-CovSel) for selecting generalized features across domains

Cited by 7 publications

References 36 publications

Partial least squares regression with multiple domains

Partial least squares regression with multiple domains

Domain Adaptation for In-Line Allergen Classification of Agri-Food Powders Using Near-Infrared Spectroscopy

A Federated Classification Approach of Waste Lubricant Oils in Geographically Distributed Laboratories

Contact Info

Product

Resources

About