2 ms·
I think it is because entropy of data is usually very high, and it is very hard to predict which architecture will work for your data the best without trying.
by riku_iki 9y ago
I think it is because entropy of data is usually very high, and it is very hard to predict which architecture will work for your data the best without trying.