In this chapter, we have seen the characteristics of over fitting and how they can be handled through L1 and L2 regularizations, and dropout. Similarly, we have seen the scenario where there was quite a lot of underfitting and how scaling or batch normalization helped us in improving the under-fitting scenario.