high bias ( under fitting): fixed by trying a larger of features,
high variance (overfitting): fixed by trying a smaller of features, including more training examples.
The bias increases as penalty parameter of the regularization increases.
The variance decreases as penalty parameter of the regularization increase.