Today I learned about comparing your models result to human level results and then improving your models on basis of human-level performance.
Then I learned about how to reduce avoidable bias and variance using several techniques which can boost up your models performance.
I also Learned about manuall error analysis and learned about several condtions where error analysis can be helpful.
Learned about how to clean mislabbeld images in dataset, how to add valueable information instead of that mislabbeled images.
I am on a break from some times using some Modern framework like TensorFlow or Keras but thinkning of starting a kaggle competiton where I will use all the skills I have learned so far in practical manner. I will start from bottom like titanic data set and aim to become a Kaggle grandmaster Inshallah.!