Gang Fang's Blog

Notes

• I have been studying statistical learning theory and particularly the relationship between the generalization gap and model capacity and amount of data. The theories are expressed in math and I realized they don’t really show up especially during practice. Their implications are simple and so during practice people simply take them for granted and rely on the implications. Hmm, I guess that is how it happens for most engineering.. We don’t try to prove the theorem all over again.