Machine Learning

Wait! machine learning?

Yes, exactly. We deal with tons of data, finding pattern from geophysical observations, retrieving information from seismic waves, earthquake detection and earthquake early warning.

It is cool. "Artificial intelligence will shape our future more powerfully than any other innovation this century. Anyone who does not understand it will soon find themselves feeling left behind, waking up in a world full of technology that feels more and more like magic." (a quote from Machine Learning for Humans, Vishal Maini & Samer Sabri)

How machine learning works?

Machine learning is basically a conceptual technique that simulate the human brain. Like how we recognize a cat or dog, you brain is a meticulous machine that process all the information from eye's input and mouth's output. For computer, scientists can simulate this process by pixels matching, compare this input image with tons of pre-build image and finding the most possible answer. This process can be improved by more complicated calculation or layers, such as convolution neural network (CNN) that reformatting the initial input pixel by small window feature matching (convolution), pooling that reduce the sensitivity spatially and rectified linear units (ReLU) that ignore the negative points. With applying these techniques, computer is, amazingly smart!

But wait, how about the unknown things?

Machine learning is not just for image recognization, it can also classify things without given labels (answers). That makes machine learning super powerful because one doesn't need to be the "supervisor" anymore, and the computer can learn itself. This is called the "unsupervised" machine learning that the computer can find patterns without given a known feature. That means you don't need to program it to learn how to understand the rule of seismic waves and picking P/S waves or what governs the rupture physics.