Explainable AI for unveiling deep learning pollen classification model based on fusion of scattered light patterns and fluorescence spectroscopy
Аутори
Brdar, SanjaPanic, Marko
Matavulj, Predrag
Stanković, Mira
Bartolić, Dragana
Sikoparija, Branko
Чланак у часопису (Објављена верзија)
Метаподаци
Приказ свих података о документуАпстракт
Pollen monitoring have become data-intensive in recent years as real-time detectors are deployed
to classify airborne pollen grains. Machine learning models with a focus on deep learning, have
an essential role in the pollen classifcation task. Within this study we developed an explainable
framework to unveil a deep learning model for pollen classifcation. Model works on data coming
from single particle detector (Rapid-E) that records for each particle optical fngerprint with scattered
light and laser induced fuorescence. Morphological properties of a particle are sensed with the
light scattering process, while chemical properties are encoded with fuorescence spectrum and
fuorescence lifetime induced by high-resolution laser. By utilizing these three data modalities,
scattering, spectrum, and lifetime, deep learning-based models with millions of parameters are
learned to distinguish diferent pollen classes, but a proper understanding of such a black-box model
decisions demand...s additional methods to employ. Our study provides the frst results of applied
explainable artifcial intelligence (xAI) methodology on the pollen classifcation model. Extracted
knowledge on the important features that attribute to the predicting particular pollen classes is
further examined from the perspective of domain knowledge and compared to available reference
data on pollen sizes, shape, and laboratory spectrofuorometer measurements.
Кључне речи:
pollen classification / automatic monitoring / fluorescence / light scattering / deep learning / explainable artificial intelligenceИзвор:
Scientific Reports, 2023, 13, 3205Издавач:
- Springer Nature
Финансирање / пројекти:
- BREATHE - Real-Time Detection and Quantification of Bioaerosols Relevant for Human and Plant Health (RS-6039613)
- Министарство науке, технолошког развоја и иновација Републике Србије, институционално финансирање - 200358 (BioSense институт) (RS-200358)
Институција/група
Institut za multidisciplinarna istraživanjaTY - JOUR AU - Brdar, Sanja AU - Panic, Marko AU - Matavulj, Predrag AU - Stanković, Mira AU - Bartolić, Dragana AU - Sikoparija, Branko PY - 2023 UR - http://rimsi.imsi.bg.ac.rs/handle/123456789/1799 AB - Pollen monitoring have become data-intensive in recent years as real-time detectors are deployed to classify airborne pollen grains. Machine learning models with a focus on deep learning, have an essential role in the pollen classifcation task. Within this study we developed an explainable framework to unveil a deep learning model for pollen classifcation. Model works on data coming from single particle detector (Rapid-E) that records for each particle optical fngerprint with scattered light and laser induced fuorescence. Morphological properties of a particle are sensed with the light scattering process, while chemical properties are encoded with fuorescence spectrum and fuorescence lifetime induced by high-resolution laser. By utilizing these three data modalities, scattering, spectrum, and lifetime, deep learning-based models with millions of parameters are learned to distinguish diferent pollen classes, but a proper understanding of such a black-box model decisions demands additional methods to employ. Our study provides the frst results of applied explainable artifcial intelligence (xAI) methodology on the pollen classifcation model. Extracted knowledge on the important features that attribute to the predicting particular pollen classes is further examined from the perspective of domain knowledge and compared to available reference data on pollen sizes, shape, and laboratory spectrofuorometer measurements. PB - Springer Nature T2 - Scientific Reports T1 - Explainable AI for unveiling deep learning pollen classification model based on fusion of scattered light patterns and fluorescence spectroscopy IS - 3205 VL - 13 DO - 10.1038/s41598-023-30064-6 ER -
@article{ author = "Brdar, Sanja and Panic, Marko and Matavulj, Predrag and Stanković, Mira and Bartolić, Dragana and Sikoparija, Branko", year = "2023", abstract = "Pollen monitoring have become data-intensive in recent years as real-time detectors are deployed to classify airborne pollen grains. Machine learning models with a focus on deep learning, have an essential role in the pollen classifcation task. Within this study we developed an explainable framework to unveil a deep learning model for pollen classifcation. Model works on data coming from single particle detector (Rapid-E) that records for each particle optical fngerprint with scattered light and laser induced fuorescence. Morphological properties of a particle are sensed with the light scattering process, while chemical properties are encoded with fuorescence spectrum and fuorescence lifetime induced by high-resolution laser. By utilizing these three data modalities, scattering, spectrum, and lifetime, deep learning-based models with millions of parameters are learned to distinguish diferent pollen classes, but a proper understanding of such a black-box model decisions demands additional methods to employ. Our study provides the frst results of applied explainable artifcial intelligence (xAI) methodology on the pollen classifcation model. Extracted knowledge on the important features that attribute to the predicting particular pollen classes is further examined from the perspective of domain knowledge and compared to available reference data on pollen sizes, shape, and laboratory spectrofuorometer measurements.", publisher = "Springer Nature", journal = "Scientific Reports", title = "Explainable AI for unveiling deep learning pollen classification model based on fusion of scattered light patterns and fluorescence spectroscopy", number = "3205", volume = "13", doi = "10.1038/s41598-023-30064-6" }
Brdar, S., Panic, M., Matavulj, P., Stanković, M., Bartolić, D.,& Sikoparija, B.. (2023). Explainable AI for unveiling deep learning pollen classification model based on fusion of scattered light patterns and fluorescence spectroscopy. in Scientific Reports Springer Nature., 13(3205). https://doi.org/10.1038/s41598-023-30064-6
Brdar S, Panic M, Matavulj P, Stanković M, Bartolić D, Sikoparija B. Explainable AI for unveiling deep learning pollen classification model based on fusion of scattered light patterns and fluorescence spectroscopy. in Scientific Reports. 2023;13(3205). doi:10.1038/s41598-023-30064-6 .
Brdar, Sanja, Panic, Marko, Matavulj, Predrag, Stanković, Mira, Bartolić, Dragana, Sikoparija, Branko, "Explainable AI for unveiling deep learning pollen classification model based on fusion of scattered light patterns and fluorescence spectroscopy" in Scientific Reports, 13, no. 3205 (2023), https://doi.org/10.1038/s41598-023-30064-6 . .