Correction to “Shapley Additive exPlanations-Integrated Convolutional Neural Networks for Chemically Interpretable Fourier-Transform Infrared-Based Microplastic Characterization”

Shapley Additive exPlanations-Integrated Convolutional Neural Networks for Chemically Interpretable Fourier-Transform Infrared-Based Microplastic Characterization

AI summary · Read the abstract

A CNN-SHAP fusion framework achieved 99.6% accuracy in classifying six common plastic polymers from FTIR spectra while using SHAP attribution to identify the specific spectral features driving each classification decision. Making microplastic identification explainable and chemically interpretable is essential for building scientific confidence in automated monitoring systems used to track plastic pollution in environmental samples.

Explainable Detection of Microplastics Using Transformer Neural Networks

AI summary · Read the abstract

Transformer neural network models can detect and classify microplastics in images with high accuracy while providing explainable outputs that identify which visual features drive classification decisions. Automated detection tools like this are key to scaling up microplastic monitoring programs and generating the standardized data needed to quantify pollution trends globally.

Deep convolutional neural networks for aged microplastics identification by Fourier transform infrared spectra classification

AI summary · Read the abstract

This study developed a deep learning model using convolutional neural networks to automatically identify aged microplastics from their infrared spectra. Aging changes the chemical signature of plastics, making them harder to identify with conventional spectral databases. The AI approach achieved high accuracy and could significantly speed up the analysis of environmental samples where weathered microplastics are the norm.

Unveiling uncertainty in microplastic quantification: Artificial intelligence integrated Raman analysis

AI summary

AI-integrated Raman analysis shows heterogeneous real-world MP samples require 40–70% filter coverage for accurate count extrapolation; frequent small-block sampling is most efficient.

Enhancing microplastic classification through filter-interfered FTIR spectra using dimensionality reduction and deep learning in low-dimensional spaces