*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Atlanta, GA | Posted: January 19, 2022
School of Electrical and Computer Engineering
Summary Sentence:
Huiye Liu and Douglas Blough received the Best Paper Award at the IEEE Consumer Communications & Networking Conference, which was held January 8-11, 2022 in a virtual format.
Full Summary:
Huiye Liu and Douglas Blough received the Best Paper Award at the IEEE Consumer Communications & Networking Conference, which was held January 8-11, 2022 in a virtual format.
Huiye Liu and Douglas BloughHuiye Liu and Douglas Blough received the Best Paper Award at the IEEE Consumer Communications & Networking Conference, which was held January 8-11, 2022 in a virtual format.
Liu is a Ph.D. student in the Georgia Tech School of Electrical and Computer Engineering (ECE). She is a member of the Critical Networking Laboratory, which is led by Blough. He is a professor and associate chair for Faculty Development in the School.
The title of their award-winning paper is “Cooperative Task-Oriented Group Formation for Vehicular Networks.” While machine learning has brought exciting advances to connected and autonomous vehicles, to date this has been done primarily through centralized batch learning, which requires very large datasets with highly accurate labels that are extremely labor-intensive to produce. Liu’s and Blough’s research has proposed a decentralized active learning approach, where vehicles collaborate with each other to create accurate labels automatically by leveraging multiple perspectives of the same scene and to learn on the fly as vehicles encounter new information not represented in their existing models.
In this paper, Liu and Blough present a novel approach for forming vehicle groups to run collaborative learning tasks, where the group structure is tailored to the specific characteristics of the tasks to be performed. The paper describes a generic distributed learning application that they developed in C++ and Python running on top of the popular Veins/SUMO vehicular networking simulation environment. The implementation was used to evaluate their task-oriented group formation approach and compare it against existing task-agnostic group formation techniques. The results demonstrate that their task-oriented approach constructs larger groups that produce better learning results while at the same time achieving a significantly higher task completion rate, as compared to task-agnostic techniques.
