|
ÄÚµå¹øÈ£ : 85 |
|
¹ßÇ¥ÀÚ : ¼ÒÁøÇö |
|
¼Ò¼Ó : DGIST |
|
ºÎ¼ : |
|
Á÷À§ : ±³¼ö |
|
¼¼¼Ç½Ã°£ : 13:00~14:50 |
|
¹ßÇ¥ÀÚ¾à·Â : |
- Assistant Professor, EECS, DGIST (2024~)
- Staff Research Engineer, Samsung Cellular & Multimedia Labs, San Diego, USA (2022-2024)
- Senior Engineer, Modem Development Team, S.LSI, Samsung, South Korea (2013-2017)
- PhD. ECE, University of Southern California, USA (2022)
- M.S., EE, KAIST (2012)
- B.S., EE, KAIST (2010) |
|
|
°¿¬¿ä¾à : |
Large-scale deployments of low Earth orbit (LEO) satellites collect massive amount of Earth imageries and sensor data, which can empower machine learning (ML) to address global challenges such as real-time disaster navigation and mitigation. In this talk, I show fundamental challenges in applying existing federated learning (FL) algorithms among satellites and ground stations, and then introduce a new FL framework which dynamically schedules model aggregation based on the deterministic and time-varying connectivity according to satellite orbits and location of ground stations. |
|
|
|