SIIM ISIC Melanoma Classification using ResNet50 Transfer learning

An implementation of model classifying Melanoma (malignant) against Benign conditions using ResNet50 Image classification model. Work under progress

Find the dataset here

The dataset is a 1024x1024 sized image with around 98% records being Benign and rest being Malignant.
The dataset is huge with around 33k high resolution images and harder to train on.
Made use of the Kaggle's TPU facility to reduce the computation complexity.

More layer is better but because of the Vanishing gradient problem model weights of the first layer cannot be updated correctly through the backpropogation of the error gradient (the chain rule multiplies the error gradient values lower than 1 and by the time gradient error reaches the first layers, its value goes to zero
That is the objective of Resnet : preserve the gradient.
How ? Thanks to the Idendity matrix because “what if we were to backpropagate through the identity function? Then the gradient would simply be multiplied by 1 and nothing would happen to it!”.

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
README.md		README.md
melanoma detection using InceptionV3.ipynb		melanoma detection using InceptionV3.ipynb
melanoma-detection-using-inceptionv3.ipynb		melanoma-detection-using-inceptionv3.ipynb
melanoma-detection-using-resnet50.ipynb		melanoma-detection-using-resnet50.ipynb
melanoma_detection.ipynb		melanoma_detection.ipynb
transfer-learning-using-resnet50 (1).ipynb		transfer-learning-using-resnet50 (1).ipynb
transfer-learning-using-resnet50.ipynb		transfer-learning-using-resnet50.ipynb

Provide feedback