Coresets are small, weighted datasets that can be used as proxies of the original dataset in performing machine learning tasks. While previous coreset construction algorithms are tailor-made for specific machine learning tasks, we proved that k-clustering based coreset provides guaranteed performance for all machine learning tasks with sufficiently continuous cost functions. This is the first […]
Our first work on using Markov Decision Process (MDP) to model dynamic service placement in mobile edge computing systems was finally accepted to IEEE/ACM Trans. Networking! The work was done in 2015, when Dr. Shiqiang Wang was still Mr. Shiqiang Wang and an intern at IBM. I am debating whether the lesson is to be […]
Yilei and Vaji had a fun time with other attendants from the DAIS community.
To present her first paper (!) on “Looking Glass of NFV: Inferring the Structure and State of NFV Network from External Observations“. Congrats to Yilei!
In Penn State Today (March 4, 2019), the top story titled “Researchers look for successful end to power grid failures” is about our newly started NSF project Coupled cAscade Modeling, Prevention, and Recovery (CAMPR)!
This paper complements our INFOCOM’19 paper on NFV Topology Inference, by focusing on a subproblem called “weight inference problem” that infers the categories of edges in the ground truth topology and the sum edge weight in each category, both from end-to-end multicast measurements. Congrats to Yilei for your second paper and keep up the good […]
Our paper “Adaptive Federated Learning in Resource Constrained Edge Computing Systems” has been accepted to the Special Issue of IEEE JSAC on Artificial Intelligence and Machine Learning for Networking and Communications! This is the journal version of our earlier INFOCOM paper, in collaboration with Dr. Shiqiang Wang, Dr. Theodoros Salonidis, and Dr. Christian Makaya at […]
Yilei’s paper on “Looking Glass of NFV: Inferring the Structure and State of NFV Network from External Observation”, Vaji’s paper on “Service Placement and Request Scheduling for Data-intensive Applications in Edge Clouds“, and our collaborator Stephen’s work on “Service Placement with Provable Guarantees in Heterogeneous Edge Computing Systems“, all funded by DAIS ITA. Congratulations to […]
The paper considers a hot problem, fast rule updates, in the context of Software-Defined Networking (SDN) from a novel perspective of path selection. The solution essentially load-balances among switches to minimize update time at the straggler switch. See details at TNSM’18.
At the DAIS ITA Annual Fall Meeting this week (9/10-12/2018), we presented two papers and three posters based on our ITA-funded projects and related projects. These are based on four long papers currently under submission, so stay tuned. First-time formal presentations for Hanlin and Yilei, good job! Besides interesting discussions, we also had a great […]