Chi-Hsien (Eric) Yen

About

Hi! my name is Eric. I am a PhD student studying Computer Science at the University of Illinois at Urbana-Champaign, supervised by Dr. Wai-Tat Fu.

My research interests lie in the intersection of human-computer interaction and data analytics. My research currently focuses on understanding and assisting human causal reasoning in visual analytics.

  • Email: cyen4 (at) illinois (dot) edu
  • CV: [PDF]

Research

Causal Inference in Visual Analytics


Open data and easy-to-use visualization tools are increasingly available to the general public; however, little is known about how, and how well, people use visualizations to infer causality. Spurious correlations and conflicting visual patterns could confuse novices and lead them to make incorrect causal inferences. We conduct user studies to understand common human reasoning strategies and pitfalls, and design visual analytics interfaces that assist causal inference.

Charitable Crowdfunding Platform


The funding distribution in existing crowdfunding platforms is often skewed: very few campaigns receive most of the money while the majority of campaigns receive little. Inspired by the Deferred Acceptance algorithm, we designed a new donating interface where donors can express their preferences over multiple campaigns. Our platform then automatically allocates funding based on donors' preferences in a more effective way, thus increasing the overall success rate.

BioScope

BioScope is a bandage-like extensible sensing system, which is specially designed for nurses. The system is easy to apply and is able to collect numerous crucial bio-signals such as ECG, temperature, movements, and sounds. I utilized 3D printing for prototyping and help conducted experiments of the system.

  • Paper: UbiComp'14 [PDF]

SoberDiary


We designed a smartphone application, SoberDiary, that helps alcohol dependent patients stay sober after medical treatments. SoberDiary consists of (1) a portable Bluetooth breathalyzer, (2) a phone application, and (3) a backend server.

Patients could take breathe alcohol content test any time and anywhere. Based on their test results and life experience feedbacks, SoberDiary automatically provides useful suggestions to help patients stay healthy, and helps doctors understand the immediate situations of the patients. I participated the user study that recruited real patients for our experiment, conducted interviews, and visualized collected data for analysis.


Wireless Protocol Design



We analyzed the channel utilization of Common Control Channel (CCC)-based wireless protocol. An efficient estimation algorithm for the number of devices on a wireless channel is designed to optimize the utilization based on our analysis. Our proposed protocol outperforms other protocols significantly, especially when the number of devices is large and in a state of dramatic fluctuation. Our work was accepted and presented in IWCMC 2013, and was selected for publication in the WCMC journal. Further improvement of the protocol is achieved by adopting the asynchronous resource reservation concept.

The figure above shows an example of the channel reservation result by our extended protocol. 20 machines, labeled from 0 to 19, are communicating on 1 CCC and 4 data channels. The length and arrival time of each packets are randomized to simulate real-world scenarios. The result shows that the data channels have been fully utilized and thus optimized the utilization.

  • Paper: IWCMC 2013 [PDF]
  • Journal: WCMC (2015): 1015-1025. [PDF]

Fun Projects

TDTOS - T-shirt Design and Try On System



We designed a T-shirts try-on simulation system which is implemented on an FPGA board. An image processing algorithm is develeoped to achieve real-time and accurate performance. We also provided an intuitive user-interface which allows the users to try on different colors efficiently by using a simple touch panel, and change T-shirt designs with hand gestures. Furthermore, users can design their own T-shirt by taking a picture of their inpromptu drawing through our camera, and place it onto the T-shirt with any size and anywhere using three 3M-brand stickers. Our system was presented in the Innovate Asia Altera Design Contest in 2012, and was awarded first prize out of 412 competing teams from asia.


Facebook Tetris Battle AI



This is a final project, Tetris AI, from Artificial Intelligence course. Many have studied about how many rows can be eliminated at most in a Tetris game. However, in real world, Tetris game players will get extra bonus by eliminating rows simultaneously or consecutively (i.e., Combo). We developed an AI that utilized the B2B Tetris strategy, which means every elimination contains 4 rows, and deployed it to real world Facebook Tetris Battle game. The demo video shows how the proposed AI sent 104 lines in 2 minutes and defeated the human player with 3 K.O..

  • Final Project Paper [PDF]