Classification of challenging Laser-Induced Breakdown Spectroscopy soil sample data - EMSLIBS contest

Vrábel, Jakub; Képeš, Erik; Duponchel, Ludovic; Motto-Ros, Vincent; Fabre, C.; Connemann, Sven; Schreckenberg, Frederik; Prasse, Paul; Riebe, Daniel; Junjuri, Rajendhar; Gundawar, Manoj Kumar; Tan, Xiaofeng; Pořízka, Pavel; Kaiser, Jozef

doi:10.1016/j.sab.2020.105872

Cited by 52 publications

(16 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, 20000 and 18000 training spectra lead to similar accuracy for this test set. MLP-LIBS-20000i shows an average accuracy of 88.2% that is similar to that of the other ML models developed for this dataset (Vrabel et al, 2020), which however do not support adaptation. Our MLP-LIBS-ADAPT using 20000L-100U shows 1.1% average improvement over MLP-LIBS, and 1.2% average improvement when using 18000L-100U.…”

Section: Model Adaptation Evaluationmentioning

confidence: 68%

“…We first introduce a new lightweight multi-layer perceptron (MLP) model, called MLP-LIBS, that achieves an average accuracy of 88.2% on a well-known LIBS dataset with 12 The MLP-LIBS model consists of only two hidden layers with 64 neurons each. MLP-LIBS's accuracy is on par with the other ML models that have been introduced in a recent work for this dataset (Vrabel et al, 2020), which however do not handle domain shift.…”

Section: Introductionmentioning

confidence: 85%

“…ML has been shown to be effective in handling high-dimensional and complex LIBS spectra, eliminating interference and noise, and improving the classification accuracy compared to other common statistical methods (Chen et al, 2020). A variety of ML models have been used: random forests, support vector machines, as well as neural networks (Pořízka et al, 2017;Chen et al, 2020;Vrabel et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Semi-supervised on-device neural network adaptation for remote and portable laser-induced breakdown spectroscopy

Bhardwaj,

Gokhale

2021

Preprint

View full text Add to dashboard Cite

Laser-induced breakdown spectroscopy (LIBS) is a popular, fast elemental analysis technique used to determine the chemical composition of target samples, such as in industrial analysis of metals or in space exploration. Recently, there has been a rise in the use of machine learning (ML) techniques for LIBS data processing. However, ML for LIBS is challenging as: (i) the predictive models must be lightweight since they need to be deployed in highly resource-constrained and battery-operated portable LIBS systems; and (ii) since these systems can be remote, the models must be able to self-adapt to any domain shift in input distributions which could be due to the lack of different types of inputs in training data or dynamic environmental/sensor noise. This on-device retraining of model should not only be fast but also unsupervised due to the absence of new labeled data in remote LIBS systems. We introduce a lightweight multi-layer perceptron (MLP) model for LIBS that can be adapted on-device without requiring labels for new input data. It shows 89.3% average accuracy during data streaming, and up to 2.1% better accuracy compared to an MLP model that does not support adaptation. Finally, we also characterize the inference and retraining performance of our model on Google Pixel2 phone.

show abstract

Section: Model Adaptation Evaluationmentioning

confidence: 68%

Section: Introductionmentioning

confidence: 85%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Semi-supervised on-device neural network adaptation for remote and portable laser-induced breakdown spectroscopy

Bhardwaj,

Gokhale

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Sun et al 25 developed a back-propagation neural network (BPNN) to predict the trace elements in soils. Tan et al 26 used an articial neural network (ANN) model to classify 70 000 spectra into 12 types of soil with an accuracy of 76.96%.…”

Section: Introductionmentioning

confidence: 99%

Simultaneous determination of lithology and major elements in rocks using laser-induced breakdown spectroscopy (LIBS) coupled with a deep convolutional neural network

Chen

Pei

Pisonero

et al. 2022

J. Anal. At. Spectrom.

View full text Add to dashboard Cite

show abstract

“…In that sense, the first LIBS benchmark dataset has been published by Képeš et al [ 13 ] which contains LIBS spectra from 138 soil samples belonging to 12 classes. Moreover, based on this benchmark dataset, a comparative classification contest has been performed during the EMSLIBS 2019 conference by Vrábel et al [ 14 ].…”

Section: Introductionmentioning

confidence: 99%

Laser-Induced Breakdown Spectroscopy: An Efficient Tool for Food Science and Technology (from the Analysis of Martian Rocks to the Analysis of Olive Oil, Honey, Milk, and Other Natural Earth Products)

et al. 2021

View full text Add to dashboard Cite

Laser-Induced Breakdown Spectroscopy (LIBS), having reached a level of maturity during the last few years, is generally considered as a very powerful and efficient analytical tool, and it has been proposed for a broad range of applications, extending from space exploration down to terrestrial applications, from cultural heritage to food science and security. Over the last decade, there has been a rapidly growing sub-field concerning the application of LIBS for food analysis, safety, and security, which along with the implementation of machine learning and chemometric algorithms opens new perspectives and possibilities. The present review intends to provide a short overview of the current state-of-the-art research activities concerning the application of LIBS for the analysis of foodstuffs, with the emphasis given to olive oil, honey, and milk.

show abstract

Classification of challenging Laser-Induced Breakdown Spectroscopy soil sample data - EMSLIBS contest

Cited by 52 publications

References 20 publications

Semi-supervised on-device neural network adaptation for remote and portable laser-induced breakdown spectroscopy

Semi-supervised on-device neural network adaptation for remote and portable laser-induced breakdown spectroscopy

Simultaneous determination of lithology and major elements in rocks using laser-induced breakdown spectroscopy (LIBS) coupled with a deep convolutional neural network

Laser-Induced Breakdown Spectroscopy: An Efficient Tool for Food Science and Technology (from the Analysis of Martian Rocks to the Analysis of Olive Oil, Honey, Milk, and Other Natural Earth Products)

Contact Info

Product

Resources

About