## Thursday, May 25, 2017

### my commentary of Calculus

For my first time learning the theory & calculation of Neural Network, I was shocked by its fundamental logic of arithmetic--Calculus.

Calculus is an sophisticated abstract of human wisdom among at least three hundrad years. The first concept of Calculus was invented by Newton, when he was pondering that can people be able to describe the movement of a dot on a curve line by using a linear function? The answer is for sure, yes. Indeed, all of the concepts of calculus that we learn today are absolutely based on the question which Newton asked in hundreds years ago.

In my opinion, Calculus is the most important arithmetic in field of Artificial Intelligence. In some perspective, Google's Alphago cannot be developed without Calculus. Let me take neural network as example, when a super complex logic program is developed, it definitely involves a huge of amount of calculation which the CPU has to deal with. However, if this program is not optimized, ultimately, the CPU will be overloaded. As I mentioned before, Calculus is a description of movement, hence by involving calculus arithmetic in the program, it is much easier for developers to find the best way of optimization.

Beyond all doubt, I was literally shocked by its precise logic when I was studying Calculus on Youtube. Believe or not, the standard calculus equation provided below is actually similar to the equation of slope that we learned in Algebra. Compared to Algebra, Calculus has an additional precise concept which is limit. People want to use 0 as denominator to express a movement; however, the basic knowledge tells us that denominator cannot be 0. Thus the smart scientists use limit to express that the line is always in movement approaching to its final distention 0.