Time: Apr, 2019 - Jun, 2019
Descriptions: The goal is to generate piano music. We downloaded 3-hour piano music, transformed them into pictures, and used them to train the PixelCNN, a generative network. After that, we used the PixelCNN to generate several segments of the music. Then K-means method was used to cluster music of the same genre and we concatenated the fragments in the same genre to get the complete music.
Time: Apr, 2019 - Jun, 2019
Descriptions: In a query process, people prefer to get the information they need with as few questions as possible. The system I designed used Bayesian Network to incorporate empirical knowledge and search for the best questions to be asked, which is most likely to give us as much information as possible. The system was tested on the toy math test project and showed expected results.
Time: Apr, 2019 - Jun, 2019
Descriptios: I led a group designed and finished a finite element software. The software has about ten different types of elements and has parallel computations post processing and other advanced functions in it. Users can input a legal txt-input and the software can solve the problem and show the deformed picture of it. Finally, we used the software we coded to design a bridge.