Relative Content

Tag Archive for pythonmachine-learningdeep-learningneural-networkcomputer-vision

Why is my validation accuracy so bad than train accuracy?

I am training ResNet50 on Chestxray dataset. My training data is around 15000 Samples, 1500 in validation and 3000 in testing. It is multilabel classification problem and I have 13 labels and I have one hot encoded it. I’m encountering a issue while training. Despite observing an improvement in training accuracy over epochs, the validation accuracy seems to degrade. I’ve tried several experiments (manipulate learning rate, change architecture, )and troubleshooting steps but haven’t been able to resolve this issue. I have used alot of data augmentations too but it is not converging. What should i do to sort it ?