Time and Frequency

20 years ago I majored my bachelor in Adaptive learning. I realize now, that’s actually the most basic idea of AI (Artificial Intelligence).

Time is what is happening in our life. We do this, we do that. It’s also what is happening to price, index, etc. We could easily transform time into frequency, for example is using a Fourier Transform or simply use Fast Fourier Transform (FFT).

source: wikipedia

A very basic formula of time is

T (period) = 1 / f (frequency)

We agree on this very simple basic math formula. It means we CAN transform something between them, vice versa.

Here comes learning into play. Once the time is transformed into frequency, we could easily see a pattern. In a most simple case, a continuous noise sound at x frequency will only look like a single x dot/line in frequency. It’s then being fed into learning algorithm and we found a pattern. Amazingly I could reduce up to 80dB (decible) in real time using technology that time (i386 processor) in a very easy way. In today’s world, we wear this noise cancellation, almost everywhere.

When life is going more complicated, the frequency is becoming more complex and thus require more advanced learning.

With their module being shared freely, we can have a 5 years old person to operate. We don’t need to become master of microconductor to become an artist of computer, to create many arts like app, robot, etc. Using the same logic, we don’t need to master the learning formula to become an artist of AI. Today, we can do much more complicated learning pattern without having to know how it mathematically works. A three years old kid with no English-speaking parents or friends, learning English from digital media since 6 months old, can speak English fluently in his/her grammar and vocabulary without any knowledge yet on how to write and spell. Many developers know less about how computer works nor they do know how machine language and all other people modules are working, but they can use them to develop so many arts already.

Here comes our not so complicated aging finance industry. When an evidence is repeated in time series, AI can learn the pattern easily and then can do better prediction. Thus the response time to make profit is getting shorter. The good thing is, finance industry is not doing random signals because they are managed by same parties who try to maintain their wealth. Indeed central banks decisions are based on multi years of historical facts/evidences and is afraid enough to do a new experiment to the market. Therefore index/price movement IS DEFINITELY NOT random.

It then makes it harder for big fund to maintain their wealth. Therefore they should keep introducing a new thing and work harder to lead to where human evolution will be, in order to protect their wealth.

I was expecting USDCNY to run a stable YUAN in January 2019 for quite some time but it seems it run much shorter than my expectation.

Again, technology is running faster and faster and they learn faster. The wealth creator should keep evolving or else they will lead to their extinction and policy barrier is their only last hope.

Theoretically, if we can bend the frequency or feed new frequencies, and transform it back into time, it may not be a time machine but it could possibly bend our future?

image
source: marvel

my money theory

]]>