MetaNChemo: A meta-heuristic neural-based framework for chemometric analysis

Antonini, Mattia; Gaiardo, Andrea; Vecchio, Massimo

doi:10.1016/j.asoc.2020.106712

Cited by 4 publications

(3 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Environmental parameters such as relative humidity (RH), temperature (T) [45], [46] or wind speed (WS) [21] are often included in the training process. Some studies carry out such calibration in indoor environments as well [47], [48]. Commonly, the target pollutants are CO, CO2, NO, NO2, O3, SO2 and particulate matter PM1, PM2.5 and PM10.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O₃ Sensors

Tondini,

Scilla,

Casari

2024

IEEE Sensors J.

View full text Add to dashboard Cite

Low-cost sensors (LCSs) show a huge potential towards enabling the pervasive and continuous monitoring of crucial environmental parameters, supporting environment preservation and informing citizens' well-being through ubiquitous air quality data. The main drawback of LCSs is that their data is usually biased, even if LCSs are calibrated by their manufacturer at production time. More accurate in-field calibration methods based on machine learning (ML) and neural networks (NNs) are being considered in some recent studies. They typically imply LCSs co-location with reference measurement stations certified by environmental agencies. Due to seasonality effects, however, the correlation between LCSs and their reference may rapidly degrade once the LCSs are moved from the calibration site, making even really accurate calibrations useless. In this work, we specifically target this problem by optimizing the training settings of the most popular ML and NN calibration models for LCSs when a sequential split schema is adopted to separate training-and test-set. Then, we assess the degradation of the calibration over time based on the R 2 score, when the splitting of the dataset between training-and test-set is different from the classical 80% -20% ratio. This method is applied to real data gathered from an O3 sensor deployed in co-location with a certified reference station for a period of 6 months. Eventually, we show that, in the case of Long-Short Term Memory NNs, using 20% of the dataset for the training is a trade-off condition that minimizes the calibration effort and still yields a robust and long-lasting calibration.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Antonini et al [48] followed a different approach based on a classification problem. Their goal was to benchmark eight CO sensors by comparing them against a discrete range of concentrations (0, 2, 5, 10, 15, 20 and 25 ppm) as a function of RH.…”

Section: Related Workmentioning

confidence: 99%

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O₃ Sensors

Tondini,

Scilla,

Casari

2024

IEEE Sensors J.

View full text Add to dashboard Cite

show abstract

“…Based on the sensing mechanism model [ 8 ], there are three reasonably effective strategies for increasing the sensitivity and selectivity of SnO 2 : nanostructure modification, heterojunction structure building, and noble metal elements doping. For instance, a slew of remarkable research works is devoted to synthesizing SnO 2 with various nano-morphologies, such as nanowires [ 9 ], nanoflower [ 10 , 11 ], nanofibers [ 12 ], nano hollow [ 13 ], 2D flakes [ 14 ], nanosphere [ 15 , 16 ], 3D mesoporous [ 17 ], hollow sphere [ 18 ], and 3D microporous spheres [ 19 , 20 ]. However, it is extremely hard to find out the mechanisms of how the morphologies changing affects the sensing properties, and the nanostructures of the materials seem extremely difficult to be designed and predicted.…”

Section: Introductionmentioning

confidence: 99%

Investigation on Sensing Performance of Highly Doped Sb/SnO2

Feng

Gaiardo

Valt

et al. 2022

Sensors

Self Cite

View full text Add to dashboard Cite

Tin dioxide (SnO2) is the most-used semiconductor for gas sensing applications. However, lack of selectivity and humidity influence limit its potential usage. Antimony (Sb) doped SnO2 showed unique electrical and chemical properties, since the introduction of Sb ions leads to the creation of a new shallow band level and of oxygen vacancies acting as donors in SnO2. Although low-doped SnO2:Sb demonstrated an improvement of the sensing performance compared to pure SnO2, there is a lack of investigation on this material. To fill this gap, we focused this work on the study of gas sensing properties of highly doped SnO2:Sb. Morphology, crystal structure and elemental composition were characterized, highlighting that Sb doping hinders SnO2 grain growth and decreases crystallinity slightly, while lattice parameters expand after the introduction of Sb ions into the SnO2 crystal. XRF and EDS confirmed the high purity of the SnO2:Sb powders, and XPS highlighted a higher Sb concentration compared to XRF and EDS results, due to a partial Sb segregation on superficial layers of Sb/SnO2. Then, the samples were exposed to different gases, highlighting a high selectivity to NO2 with a good sensitivity and a limited influence of humidity. Lastly, an interpretation of the sensing mechanism vs. NO2 was proposed.

show abstract

Multi-surrogate assisted binary particle swarm optimization algorithm and its application for feature selection

Pan

Chu

et al. 2022

Applied Soft Computing

View full text Add to dashboard Cite

MetaNChemo: A meta-heuristic neural-based framework for chemometric analysis

Cited by 4 publications

References 45 publications

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O₃ Sensors

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O₃ Sensors

Investigation on Sensing Performance of Highly Doped Sb/SnO2

Multi-surrogate assisted binary particle swarm optimization algorithm and its application for feature selection

Contact Info

Product

Resources

About

MetaNChemo: A meta-heuristic neural-based framework for chemometric analysis

Cited by 4 publications

References 45 publications

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O3 Sensors

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O3 Sensors

Investigation on Sensing Performance of Highly Doped Sb/SnO2

Multi-surrogate assisted binary particle swarm optimization algorithm and its application for feature selection

Contact Info

Product

Resources

About

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O₃ Sensors

Minimized Training of Machine Learning-Based Calibration Methods for Low-Cost O₃ Sensors