We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Reliability Testing of Machine Learning Model Prediction Capability towards Unidentifiable Microplastic Spectral Data: Triple Battery and Colorant Investigation
Summary
Machine learning models trained on 1,800 microplastic spectra were tested against challenging unidentifiable samples including battery plastics and colorant-containing particles, finding that spectral variability from additives and weathering substantially reduces prediction reliability—highlighting a key limitation for AI-based microplastic identification.
Microplastics (MPs) are exudated fragments and fibers from environmental plastic refuse in the lower millimeter to micrometer range with some intentionally created particles and degraded exudates from anthropogenic sources. These particles are often hard to identify due to extensive weathering and material heterogeneity from additives. Leveraging the ability of machine learning (ML) models, which train on features from selected polymer classes, can aid in determining particle identity despite heterogeneity. Herein, a 1,800 spectral training dataset was employed for model training (900 from µ-FTIR and 900 from µ-Raman) using data from common MP polymers (synthetically or naturally derived) and plastics. A threefold battery; synthetic data (SD), mixed synthetic data (MSD), and real-world data from an FTIR Library of Plastic Particles sourced from the Environment (FLOPP-E) and a [Raman] Spectral Library of Plastic Particles aged in the Environment (SLOPP-E); was administered for reliability. Firstly, the SD test determined subspace k-nearest neighbors (SKNN) and wide neural network (WNN) as champions (µ-FTIR and µ-Raman, respectively) with an accuracy of, 99%/100% and 98%/100% (χ2 = 31.99/69, p = .0024/ < .0001). Secondly, the MSD test served as a progenitor to the multi-class prediction from µ-Raman’s SKNN showing consistency across 5 replicates (H = .25—6.59, p = .156—.993). And thirdly, the real-world test exhibited a loss in accuracy rate with only the champion, SKNN, retaining ~ 73% and ~ 49% of the correct predictions. Investigation into the probable causes of misprediction via colorant additives led to the discovery of white and blue as the most unpredictable across both databases. Lastly, an investigation of 2 samples revealed confounding colorant additives (SKNN’s predictions) as copper phthalocyanine (C2. Blue Fiber) and a derivative from the family of diketo-pyrrolo-pyrroles (Polyester 12. Red Fiber).
Sign in to start a discussion.
More Papers Like This
Spectroscopic Identification of Environmental Microplastics
Scientists developed a machine learning classifier that identifies the chemical type of environmental microplastic samples from spectral data with over 97% accuracy, even for samples from unknown sources. Automated spectral identification tools are critical for scaling up microplastic monitoring across large environmental datasets.
Machine learning based workflow for (micro)plastic spectral reconstruction and classification
A machine learning pipeline combining two spectral reconstruction models with four classification algorithms can identify microplastic polymer types from spectral data with up to 98% accuracy on processed spectra. Applied to real environmental samples, the best model achieved 71% top-one accuracy and over 90% top-three accuracy. Automated, high-accuracy microplastic identification tools are critical for scaling up environmental monitoring and making large-scale surveys practical.
Enhancing Confidence in Microplastic Spectral Identification via Conformal Prediction
Researchers applied a machine-learning framework called conformal prediction to improve confidence in identifying microplastic types from spectral data. The approach produces a set of possible labels guaranteed to contain the true identity with a user-defined probability, addressing the lack of statistical confidence in standard hit-quality-index matching. This open-access tool could reduce the need for time-consuming manual spectral inspection and improve the reliability of microplastic quantification across studies.
Machine Learning Method for Microplastic Identification Using a Combination of Machine Learning and Raman Spectroscopy
Researchers developed a machine learning method for identifying microplastics using a combination of multiple spectroscopic techniques, improving classification accuracy beyond single-method approaches and enabling automated polymer identification.
Development of representative convolutional neural network based models for microplastic spectral identification
Researchers developed more representative convolutional neural network (CNN) models for microplastic spectral identification by training on expanded spectral databases that include greater diversity of plastic types, aging stages, secondary additives, pigments, and environmental contamination, outperforming library-search methods in classification accuracy and speed.