Electrical and Computer Engineering Master’s Students Present Their Work at the EE/CE Project Expo ’17

The Columbia EE/CE Master’s Student Projects Expo ’17 took place on December 13. Seventy-one students and 31 teams presented projects ranging from Deep Networks with Stochastic Depth to a Smart Hamper.

Students developed their projects in the Internet of Things (IoT) course, taught by Professor Fred Jiang; the Neural Networks and Deep Learning course, taught by Professor Zoran Kostic; the VLSI Hardware Architecture course, taught by Professor Mingoo Seok; the Content Distribution Networks course, taught by Adjunct Professor Anwar Walid; and the Wireless and Mobile Networking (WiMNet) lab, directed by Professor Gil Zussman.

“The projects were very diverse,” says Kostic, who directs the MS EE program. “It is especially stimulating for the students to see projects from areas other than their own.”

Judging the projects were Jiang and Kostic, as well as Professor Sharon Di from civil engineering and Jorge Ortiz, PhD, a research scientist at IBM T. J. Watson Research Center.

The first prize-winning project was Parallelizing Iris Detection Algorithms, by Erik Su and Zhendong Deng.Su says their motivation stemmed from the growing interest in the use of biometrics for human identification. Instead of applying the algorithms to each pixel of an eye image individually, through parallelization, Su and Deng applied the algorithms to multiple pixels simultaneously. One problem they encountered, says Deng, was interference caused by irrelevant factors such as eyelids and eyelashes.

The two second prize-winning projects were Trendy Wardrobe, by Siao-Ting Wang, Erik Su, and Miguel Belfort, and Anti-Theft System, by Ruimin Zhao, Xiaoxiang Zhang, and Minghao Li.

The motivation for Trendy Wardrobe was Wang’s task of selecting her daughter’s clothes for school each day. The project has two main features: a clothes-detection system that identifies and locates available garments in the wardrobe, and a clothes-recommendation system that picks the most suitable attire based on analysis of data such as the current weather, the user's preferences, and activities in the user’s calendar.

Zhao and Zhang’s innovation in designing their Anti-Theft System was to use a vibration sensor to identify a person’s gait through deep learning. The use of a vibration detector is less expensive and more energy-efficient than that of a vision-based system.

The three third-prize winners were A Neural Algorithm of Artistic Style, by Siyu Liu, Vinay Kale, and Yue Wen; Parallelized NLP Market Prediction, by Sam Beaulieu; and Parallel Implementation of GloVe, by Zixiao Zhang and Di Zuo.

The latter transforms text words into dense numerical vectors, which maintain the relations of the original words. Simple calculation can be used to describe these relations; this is especially useful in Natural Language Processing, the study of the interactions between computers and human languages.

“What is perhaps most impressive about all the projects,” says Jiang, “is that the students came up with the ideas on their own. They found ways to apply the knowledge they had gained in class to design solutions to real-world problems.”


-By Ann Rae Jonas

500 W. 120th St., Mudd 1310, New York, NY 10027    212-854-3105               
©2014 Columbia University