Am working on credit card fraud detection problem using autoencoders. Regarding that I have some doubts given below :
The dataset for the above problem has been downloaded from kaggle which is highly imbalanced. That is, only 494 frauds are there in the dataset comprising of 2,84,807 transactions. So my doubt is why the dataset is not balanced before applying autoencoder on it.
https://medium.com/@curiousily/credit-card-fraud-detection-using-autoencoders-in-keras-tensorflow-for-hackers-part-vii-20e0c85301bd I have read this blog post and my doubt is what is the threshold value for setting a boundary for anomaly detection.
Are autoencoders used for anomaly detection?