In the summer vacation,I attended three AI competitions.One of the competitions was short of one to qualify for the second round. The process ideas and codes for two of these competitions have been put on my GitHub.For the details of the competitions, you can watch code in the relevant repositories.

During the competition, I have some practical experience in the development process of AI. And through my own thinking and learning, I learned some methods of data processing, model building and parameter adjustment.Through these matches, I have gained a lot of practical experience.Of course, there are also many problems in the competition, such as how to resist over-fitting, how to make the generalization ability of the model better, how to adjust the parameters, what to adjust, whether to adjust large or small, how much to adjust, and so on.

Up to now, I have been studying AI only half a year.So the ranking of these competitions may not be very high. So I need more time to learn the knowledge of AI and some skills.

For these matches, I will pay attention to the final defense, learning how the TOP teams do data processing, how to build models, how to adjust parameters, and what tricks they use. In order to practice in the future, my ability can continue to improve and get higher ranking. A competition without open source code and process is Huawei’s garbage sorting AI competition.The final defense of the competition, I’ll watch it in scene.

There is still a long way to go to study, and I will continue to go on.