Since joining Acer, I have been involved in Android application development, primarily responsible for collecting user experience data on Acer smartphones. This includes analyzing the usage frequency of various software and log analysis. These data are uploaded to GCS and then transferred to the company's Hadoop cluster for big data analysis through Hadoop's Oozie scheduling workflow. I also designed the API interface for the backend Google App Engine. Additionally, I participated in the development of the Acer DADA project, which assists customers in performing software updates without using the Google Play Store.
Around 2019, I began delving into the field of deep learning. My work mainly involves training FaceID models, object detection models, object segmentation models, speech recognition, and designing models for emotion and age analysis.
I have particular experience in model compression; for example, when designing a dangerous shoe detection model for the Taipei Metro, I compressed the original model to run on hardware at more than half the cost while achieving better accuracy.
Moreover, I helped the team introduce and implement CI/CD processes. Improving team efficiency is one of my interests.
I am a person who loves to constantly explore, enjoys breaking existing frameworks, creating more efficient workflows, and is passionate about academic research and practical application.
Develop Acer Nidus which collects user experience on acer cellphone, analyzing the data by hadoop, and to improve the user experience
Develop a google app engine called Acer DADA, which is designed to subscribe and deploy app to acer device<Introduction> <NEWS>
Working on the computer vision program, including object detection and face recognition, and deploy the application with TensorRT or Intel OpenVINO to get acceralated
Yang, Cheng-Zen, and Ming-Hsuan Tu. "LACTA: An Enhanced Automatic Software Categorization on the Native Code of Android Applications." Proceedings of the International MultiConference of Engineers and Computer Scientists. Vol. 1. 2012.<PDF>
Ye-In Chang, Jun-Hong Shen, Chia-En Li, Zih-Siang Chen and Ming-Hsuan Tu, 2019, Oct., “Mining Image Frequent Patterns based on a Frequent Pattern List in Image Databases”, accepted by The Journal of Supercomputing (SCI). <PDF>