Chiều thứ Sáu tuần này ngày 20/01 - từ 1:30pm - 2:30pm
Địa điểm : phòng Lecture Hall - Viện JVN (Khu phố 6, Phường Linh Trung, Quận Thủ Đức - Nằm trong khuôn viên khu công nghệ phần mềm ĐHQG TP.HCM)
Topic: Feature engineering for big data
Abstract: In many real-time applications using big data, applying sophisticated models is unpractical for various reasons. They suffer longer training time, delay in adapting to environment changes and high maintenance cost. Feature engineering is a more practical approach in those situations. Not only it can help a simple linear model such as logistic regression to achieve the same performance as other sophistical models, it is pretty much a reusable process. This talk will cover the basic workflow in feature engineering, namely feature exploration, exploitation and selection. It will also discuss the framework to achieve such a result on big data.
Bio. Dung Hong (Ph.D.) has graduated from Purdue University in 2014. He then joined TellApart - a startup in California as software engineer. In 2015, TellApart was acquired by Twitter for over $500M. At TellApart and Twitter, Dung has worked on applying Machine Learning in various performance ads applications. He has led projects that drove substantial business values, from customer quality prediction to revenue optimization. Dung has recently came back to Vietnam and joined TenPoint7 as Lead Data Scientist.