# A temporal deep learning framework for calibration of low-cost air quality sensors

arXiv: https://arxiv.org/abs/2604.21527 · HTML: https://mlsanity.com/paper/2604.21527/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 46 references that are in the map.
- First posted on arXiv 23 Apr 2026.

## Builds on (most cited first, 8 of 8)

- [Deep Spatio-Temporal Residual Networks for Citywide Crowd Flows Prediction](https://mlsanity.com/paper/1610.00081/) (2016)
- [Deep Air Quality Forecasting Using Hybrid Deep Learning Framework](https://mlsanity.com/paper/1812.04783/) (2018)
- [Machine Learning for Urban Air Quality Analytics: A Survey](https://mlsanity.com/paper/2310.09620/) (2023)
- [ModelFLOWs-app: data-driven post-processing and reduced order modelling tools](https://mlsanity.com/paper/2305.17150/) (2023)
- [Few-shot calibration of low-cost air pollution (PM2.5) sensors using meta-learning](https://mlsanity.com/paper/2108.00640/) (2021)
- [On The Reliability Of Machine Learning Applications In Manufacturing Environments](https://mlsanity.com/paper/2112.06986/) (2021)
- [Blind Calibration of Air Quality Wireless Sensor Networks Using Deep Neural Networks](https://mlsanity.com/paper/2111.09417/) (2021)
- [In-field Calibration of Low-Cost Sensors through XGBoost & Aggregate Sensor Data](https://mlsanity.com/paper/2506.15840/) (2025)

## Abstract (arXiv)

Low-cost air quality sensors (LCS) provide a practical alternative to expensive regulatory-grade instruments, making dense urban monitoring networks possible. Yet their adoption is limited by calibration challenges, including sensor drift, environmental cross-sensitivity, and variability in performance from device to device. This work presents a deep learning framework for calibrating LCS measurements of PM2.5, PM10, and NO2 using a Long Short-Term Memory (LSTM) network, trained on co-located reference data from the OxAria network in Oxford, UK. Unlike the Random Forest (RF) baseline, which treats each observation independently, the proposed approach captures temporal dependencies and delayed environmental effects through sequence-based learning, achieving higher R2 values across training, validation, and test sets for all three pollutants. A feature set is constructed combining time-lagged parameters, harmonic encodings, and interaction terms to improve generalization on unseen temporal windows. Validation of unseen calibrated values against the Equivalence Spreadsheet Tool 3.1 demonstrates regulatory compliance with expanded uncertainties of 22.11% for NO2, 12.42% for PM10, and 9.1% for PM2.5.