Q$10(15$ pts$)$: Assume you need to build a model for fraud detection but you do not have target

labels which show whether that transaction is fraud or not in your dataset $($i$.$e$.$ no target label

like $Y=1$ fraud or $Y=0$ as not fraud$).$ So$,$ you may not use supervised learning. Dataset has $10$

columns $($i$.$e$.$ features$)$ and all of them are numeric. Build a deep learning model for this problem

and write the pseudo code for this model. Explain the training, test step and how to predict fraud

when new data arrives during inference.