Machine learning to find the right learning method

The fate of life, strange and difficult to test.

I thought the time was devoted to Java, but did not want to break into the hall of machine learning. That summer, the scorching sun, across 1000 kilometers to the strange city of wandering, I hope all this is worthwhile.

I Java origin, slightly understand c,linux, database, technology slag slag.

Hope every step of life is a new starting point, each step has a new state of mind.

I have never heard of this in machine learning, AI is heard, deep learning is needless to say, at the beginning of work, exposure to machine learning, and related algorithms and theoretical data formulas. To me is really a kind of suffering, those familiar things are completely not understand. Then slowly understand what the algorithm is doing. To solve any problem. With me is a bunch of graduate students, the boss completely regardless of us, although they look very busy. Show with my dream is a big gap, I imagined that there is a God general of the eldest brother to take me to fly, the result is to be put up (haha). But there is no use in saying anything, there are pros and cons. This also gives us a large part of the time to explore.

I began to develop a learning plan, collect information, watch the video, hope to understand these things in theory, slowly to practice them, and finally use. Well, it looks good. The idea is really hardships, and I finally failed to make it through the road. Theoretical knowledge involves, probability theory, Mathematical statistics, advanced mathematics. For a person who never listens to a class is really too difficult, see for a few days no longer persist. I think it is not to make up for those knowledge, and ultimately choose to give up. On the contrary, the graduate students who are with me they are able to go through a video tutorial, a little comprehension theory, over and over again every day. I only have to look at some blogs and documents every day to understand ideas. Maybe that's the difference!

Later together to study an algorithm to do a use of time, I really find how to get started, machine learning algorithm is required very rigorous theoretical basis, she does not like the usual use of some simple algorithm, just need to think about enough, all machine learning algorithms have a rigorous theory to prove his correctness. This is also the difficulty of learning, when we do not have the ability to understand the theory when there is no way to abstract into algorithmic thinking, naturally there is no way to implement an algorithm. But its features also limit its changes, so there are many libraries have implemented these algorithms, we can go into the process of the algorithm, to understand the meaning of each step, and then combined with mathematical theory to verify the rationality of these steps.

Binate theory again practice to apply again thought finally to the theory of the past, let me really feel can go to a glimpse of the mystery of machine learning. If you have the same confusion as me, you can first use the algorithm to do some small applications, to see the experimental results, in step-by-step from the code level to understand, and finally combined with the theory of mutual verification.

Machine learning to find the right learning method