TY - GEN
T1 - Optimizing Graph Neural Network Architectures with Swarm Intelligence for Improved Skin Cancer Detection
AU - Datta, Parul
AU - Padmaja, Ragolu
AU - Sahukaru, Jayavardhanarao
AU - Maram, Balajee
AU - Prasan, U. D.
AU - Sathishkumar, V. E.
N1 - Publisher Copyright:
© 2025 IEEE.
PY - 2025
Y1 - 2025
N2 - Skin cancer constitutes about 33% of all the globe's cancers, and early discovery gives survival chances of up to 98%. A new Graph Neural Network (GNN) model improved through Swarm Intelligence strategies for improved accuracy in skin cancer diagnosis is hereby presented. Deep learning methods cannot learn intricate topological relationships in lesion data, while GNNs are suitable to learn such kinds of relationships. The addition of the hybrid algorithm of Grey Wolf Optimizer (GWO) and Particle Swarm Optimization (PSO) to the GNN model enables it to deliver 7-10% improved classification accuracy in comparison with baseline GNNs. ISIC skin cancer based skin cancer dataset offers an upper limit of 94.8%, which is better than standard CNN models. The study envisions the possibility of using nature-inspired optimization and GNNs for providing robust, scalable, and interpretable medical diagnosis systems.
AB - Skin cancer constitutes about 33% of all the globe's cancers, and early discovery gives survival chances of up to 98%. A new Graph Neural Network (GNN) model improved through Swarm Intelligence strategies for improved accuracy in skin cancer diagnosis is hereby presented. Deep learning methods cannot learn intricate topological relationships in lesion data, while GNNs are suitable to learn such kinds of relationships. The addition of the hybrid algorithm of Grey Wolf Optimizer (GWO) and Particle Swarm Optimization (PSO) to the GNN model enables it to deliver 7-10% improved classification accuracy in comparison with baseline GNNs. ISIC skin cancer based skin cancer dataset offers an upper limit of 94.8%, which is better than standard CNN models. The study envisions the possibility of using nature-inspired optimization and GNNs for providing robust, scalable, and interpretable medical diagnosis systems.
KW - Graph Neural Networks
KW - Medical Image Analysis
KW - Nature-Inspired Optimization
KW - Skin Cancer Detection
KW - Swarm Intelligence
UR - https://www.scopus.com/pages/publications/105035177158
U2 - 10.1109/ICCCA66364.2025.11325637
DO - 10.1109/ICCCA66364.2025.11325637
M3 - Conference contribution
AN - SCOPUS:105035177158
T3 - 2025 IEEE 7th International Conference on Computing, Communication and Automation, ICCCA 2025
BT - 2025 IEEE 7th International Conference on Computing, Communication and Automation, ICCCA 2025
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2025 IEEE 7th International Conference on Computing, Communication and Automation, ICCCA 2025
Y2 - 28 November 2025 through 30 November 2025
ER -