Bias Correction in NWP Temperature Forecasts using LSTM-Based Deep Learning Models
S. Mandal, et al.
2025 12th International Conference on Reliability, Infocom Technologies and Optimization (ICRITO),
IEEE, Noida NCR, India. Sep 18–19, 2025.
Abstract
Numerical Weather Prediction (NWP) models are essential for accurate weather forecasting but are
often prone to systematic biases that reduce their reliability for agricultural planning and disaster
management. This paper proposes an LSTM-based deep learning framework for post-processing NWP
temperature forecasts to correct systematic biases. The model captures complex temporal patterns in
historical meteorological data to learn correction mappings. Experimental results demonstrate a
13% improvement in forecasting accuracy,
an R² score of 0.9408, and a 65% reduction in systematic bias across diverse weather conditions.
The proposed method shows significant potential for improving forecasting reliability for
7-day extended forecasts.
@inproceedings{mandal2025bias,
author = {Mandal, S. et al.},
title = {Bias Correction in NWP Temperature Forecasts using LSTM-Based Deep Learning Models},
booktitle = {2025 12th Int. Conf. on Reliability, Infocom Technologies and Optimization (ICRITO)},
year = {2025},
pages = {--},
doi = {10.1109/ICRITO66076.2025.11241393},
publisher = {IEEE}
}
Accepted — IEEE Xplore (In-Press)
2
In-PressIEEE Xplore
Feb 2026
AI and ML Approaches for Automated Mosquito Species Identification: A Review
S. Mandal, P. Singh
5th International Conference on Innovative Practices in Technology and Management (ICIPTM),
Noida NCR, India. Feb 19–20, 2026.
Abstract
Accurate identification of mosquito species is critical for controlling vector-borne diseases
such as malaria, dengue, and Zika. Manual identification is time-consuming and requires expert
entomology knowledge. This paper provides a comprehensive review of AI and machine learning
approaches for automated mosquito species identification, covering classical image processing,
CNN-based classifiers, transfer learning methods, and ensemble deep learning strategies.
We analyze existing datasets, model architectures, performance benchmarks, and open challenges,
proposing a roadmap for future research directions in automated disease vector surveillance systems.
@inproceedings{mandal2026review,
author = {Mandal, S. and Singh, P.},
title = {AI and ML Approaches for Automated Mosquito Species Identification: A Review},
booktitle = {5th Int. Conf. on Innovative Practices in Technology and Management (ICIPTM)},
year = {2026},
publisher = {IEEE}
}
3
In-PressIEEE Xplore
Feb 2026
Automated Identification of Zika Virus Mosquito Vectors Using Ensemble Deep Learning
S. Mandal, P. Singh
5th International Conference on Innovative Practices in Technology and Management (ICIPTM),
Noida NCR, India. Feb 19–20, 2026.
Abstract
Zika virus outbreaks are driven primarily by Aedes aegypti and Aedes albopictus
mosquitoes. Early and accurate identification of these vectors is critical for effective
epidemiological surveillance. This paper presents an ensemble deep learning approach combining
VGG16, ResNet50, and EfficientNet architectures for automated classification of mosquito species
from wing morphology images. The ensemble model achieves superior accuracy compared to individual
models, significantly improving robustness against image noise and class imbalance. The system
provides a scalable, low-cost tool for field deployment in Zika-endemic regions.
Zika VirusEnsemble LearningVGG16ResNet50EfficientNetMosquito VectorAedes aegypti
ICIPTM 2026
Noida NCR, India
@inproceedings{mandal2026zika,
author = {Mandal, S. and Singh, P.},
title = {Automated Identification of Zika Virus Mosquito Vectors Using Ensemble Deep Learning},
booktitle = {5th Int. Conf. on Innovative Practices in Technology and Management (ICIPTM)},
year = {2026},
publisher = {IEEE}
}