Tsinghua & Carnegie Mellon University Beijing. China& Pittsburgh. USA. 09/2018- Present
- CMU master of computer science (MSCS) & THU-CMU double degree program
Beihang University (BUAA) Beijing. China 09/2013- 07/2018
- Bachelor of Engineering in Software Engineering & Minor in English Rank 1 out of 141
- GPA: Major 3.88/4.0 & Minor 3.94/4.0
- Zhihao O. Yihang Y, Kun Y, Jian W, Xiaolin H. XiaoJoy: Using Locally Connected CNN to Cooperate with Human Composers (Artwork). Conference on Neural Information Processing System (NeurIPS)Workshop 2018.
- Zhihao O. Yan F, Zihao H, Tianbo H, Tao D, Shu-Tao X. AttentionDrop for Convolutional Neural Networks. International Conference on Multimedia and Expo (ICME)2019.
- Zhihao O, Yuyuan Z, Zihao H, Yinghua G, Tao D, Shu-Tao X. Towards Randomness in Learning Rates.International Joint Conferences on Artificial Intelligence (IJCAI)2019 (under review).
- Lu Y, Zhihao O, Xu S. HTML/CSS/JS Website Design and Development Case. Tsinghua University Press.
- Lu Y, Xu S, Zhihao O. When Deep Learning Meets Computer Vision. Tsinghua University Press (under review).
Work and Project Experience
Tsinghua University Shenzhen. China 09/2018– Present
Researcher Mentor: Prof. Shu-Tao Xia
- Randomization research. Adaptive Dropoutfor better regularization than original Dropout. Use random learning rate to bring variance to the backpropagation gradient for better optimization and enhance the model’s robustness.
- Utilized the LID (Local Intrinsic Dimensionality) to analyze the data and organize a better data distribution for training.
Tsinghua University & Deep Music Company Beijing. China 04/2018– 12/2018
Intern Researcher Mentor: Prof. Xiaolin Hu
- Used Bi-LSTM, Seq2Seq and other LSTM models for music generation.
- Designed CNN based on resNet and Locally Connected CNN structure to replace the existing RNN generation model, achieving higher precision and speed in the task of sequence generation.
Face++ Company Beijing. China 04/2018- 2/2019
- Designed face feature extraction network, using series structure such as Inception Block, Group Conv, IBNNet to improve network robustness and speed.
- The study of feature regularization, using the ideas of the NormFace series of articles, and improved the optimization objective of the classification from the perspective of the angle.
- Metric learning research. Improved the traditional triplet-loss on face recognition with n-pair loss, angular loss.
- Nist’s 2018 Face Recognition Vendor Test (FRVT) achieved the top 5 results (Megvii Team, still in progress).
Goertek Company Beijing/Shandong, China 12/2017– 04/2018
- Designed a defect detection system for the iPhone microphone production line involves the use of latest computer vision algorithms such as classification, segmentation, and detection.
- Based on DenseNet and InceptionResNet, implemented a fast classification model. With the segmentation model based on Mask-RCNN, the accuracy of 99.5% and the speed of 50 milliseconds per component were achieved. Core algorithm for a 0.2 billion $ detection system.
Microsoft Research Asia (MSRA) Beijing, China 05/2017– 08/2017
- Maintained Microsoft Face API’s customer-related code, developed face recognition related technology.
- Used LSTM to predict music chords, created interactive UI, and the relevant results were selected as the demoware in the MSRA showroom.
- AI composer MelodyMaster won Award of Most Creative in Microsoft Hackathon 2017 (2 of 12824 teams).
SoundMix Team Leader 09/2017– 04/2018
- Designed smart audio recognition software on iOS by using Fast Fourier Transform Algorithm (FFT) to realize the recognition for the pitch of instruments.
- Won the venture capital by Beijing Mousse Company.
ChordMixer Team Leader 05/2017– 08/2017
- Familiar with algorithmic composition. Generated multi-track music with machine learning. Deployed project on both Windows and Android platform.
- Designed the music chord prediction model base on the process of the HMM chain.
- Product was exhibited at the China Silk Road Science Festival. Gave speech on International Collegiate Design and Innovation Competition(ICDIC).
- Frameworks: Tensorflow, Keras, .Net, JQuery, Bootstrap.
- Additional Skills: Product Presentation, Violin, Piano, Guitar, Composition, Design, Painting.
Competitions & Awards
- Future Cup Chinese College AI challenge, second place in Beijing 06/2018
- Merit undergraduate student in Beijing 06/2018
- Champion, bestTechnology Awards(also the best design and innovation), MSRA Hackathon. 04/2018
- 1st winner of ShenYuan Medal (10 winners annually) Highest honor for students in BUAA.12/2017
- Merit Student (top 5%), 3 times, BUAA. 10/2014-10/2017
- Award of Most Creative (awarded to 2 of 14000 teams), Microsoft Global Hackathon 07/2017
- National Scholarship (awarded to top 0.2% Chinese students). 10/2016
- Second Place Internationally in Tencent International Cooperation Contest of Innovation and Entrepreneurship, (Awarded to 2 of 4000 teams). 08/2016