A Fuzzy Rank-based Deep Ensemble Methodology for Multi-Class Skin Cancer Classification

Arindam Halder, Anogh Dalal, Sanghita Gharami, Pawan Kumar Singh^{[0000-0002-9598-7981]}

Directory Structure

+-- assets/
+-- DermaMNIST_FuzzyEnsemble.ipynb
+-- GradCAM_DermaMNIST.ipynb
+-- Ham10000_FuzzyEnsemble_GradCAM.ipynb

Running Codes

Clone the project repository:

git clone https://github.com/arindam369/FuzzyEnsemble-SkinCancer.git   cd FuzzyEnsemble-SkinCancer

• Run Ham10000_FuzzyEnsemble_GradCAM.ipynb file to view the result of our fuzzy ensemble model on Ham10000 dataset.
• Run DermaMNIST_FuzzyEnsemble.ipynb file to view the result of our fuzzy ensemble model on DermaMNIST dataset.
• To visualise GradCAM analysis on DermaMNIST dataset, run the GradCAM_DermaMNIST.ipynb file.
  

assets directory contains all the images of our study including results, architectures etc.

Datasets

Two datasets are utilized in this study -

• DermaMNIST can be found at: albertvillanova/medmnist-v2 | MedMNIST
• Ham10000 can be found at: Kaggle Datasets | ISIC 2018 Challenge | Harvard Dataverse