A Federated Neuro-Symbolic Edge Intelligence Framework for Disease Prognosis and Adaptive Irrigation in Chilli Cultivation

DOI

10.2991/978-94-6239-616-6_68How to use a DOI?

Keywords

Smart Agriculture; Federated Learning; Neuro-Symbolic AI; Edge Computing; IoT Sensing; Explainable Artificial Intelligence (XAI); Disease Prognosis; Irrigation Optimization

Abstract

Smart agriculture demands scalable, privacy-preserving, and explainable intelligence to improve yield and sustainability. In this paper, AgriSensNet, a federated neuro-symbolic edge learning model, is proposed for chilli farm in the Karimnagar district of Telangana, India. The system combines distributed IoT sensing, federated deep learning, and symbolic reasoning to allow the process of disease prognosis and irrigation optimization, without centralizing the farmers sensitive data. Multi-modal IoT nodes at the sensing layer will measure soil (moisture, temperature, electrical conductivity) and plant health data by using imaging sensors. All edge nodes run a Capsule-Transformer Hybrid Network (CapFormer) to detect leaf diseases and monitor their symptoms effectively and efficiently, maintaining spatial-hierarchical connectivity that CNNs tend to eliminate. Similarly, Time series of soil and weather are meanwhile run on a Temporal Graph Convolutional Network (T-GCN) to predict inter-field interactions and irrigation requirements. Updates to the model are synchronized with the help of Federated Adaptive Optimization (FedAdaOpt) protocol balancing the global accuracy and local heterogeneity, whereas a Neuro-Symbolic Decision Layer represents the agronomic rules (e.g., irrigation thresholds, fungicide dosage) with the help of differentiable logic operators, which makes it possible to interpret the agronomic rules through the usage of artificial intelligence. Evaluation was done on a dataset of 8,200 multimodal instances of IoT deployments and regional repositories. AgriSensNet demonstrated an accuracy of 99.3% in the identification of the diseases, F1-score = 0.991, and RMSE = 0.048 in predicting the soil-moisture, higher than centralized baselines (CNN-LSTM, Hybrid-Ensemble, and XGBoost-based regressors) by a maximum of 7%. Field experiments showed a 14%-point decrease in irrigation water used and a 12%- point increase in yield, which proves that federated neuro-symbolic learning provides scalability and interpretability in the next-generation smart-farming ecosystem.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Mohd Ashfakul Hasan
AU  - K. Jagan Mohan
AU  - V. Vivekanandhan
PY  - 2026
DA  - 2026/03/31
TI  - A Federated Neuro-Symbolic Edge Intelligence Framework for Disease Prognosis and Adaptive Irrigation in Chilli Cultivation
BT  - Proceedings of the International Conference on Artificial Intelligence and Secure Data Analytics (ICAISDA 2025)
PB  - Atlantis Press
SP  - 910
EP  - 926
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-616-6_68
DO  - 10.2991/978-94-6239-616-6_68
ID  - Hasan2026
ER  -