Shay Vargaftik
Researcher
Herzliya, IL, IL
I enjoy working on interdisciplinary systems, including their design, analysis, and implementation. I am passionate about machine learning, networking, and emerging technologies.
» Can we make decision-tree-based ensemble methods accessible to edge computing? Check out our recent techniques here.
» Our new compression pipeline (EDEN) is now a part of OpenFL. See the exciting details here.
» Check out our recent efforts in pushing the limits of network efficiency for federated machine learning here.
Recent external service (program committee member):
- The 3rd Workshop on Machine Learning and Systems (EuroMLSys) '23
- USENIX Annual Technical Conference (ATC) '23
- International Conference on emerging Networking EXperiments and Technologies (CoNEXT) '22
- USENIX Annual Technical Conference (ATC) '22
- International Conference on emerging Networking EXperiments and Technologies (CoNEXT) '21
Projects
- Resource-Efficient Classification
- Efficient and Trustworthy Machine Learning
- Efficient Federated Learning
Publications
- Faster and More Accurate Measurement through Additive-Error Counters
- Virtualized Congestion Control
- dRMT: Disagreggated Programmable Switching
- Composite-Path Switching
- Memento: Making Sliding Windows Efficient for Heavy Hitters
- RADE: Resource-Efficient Supervised Anomaly Detection Using Decision Tree-Based Ensemble Methods
- C-Share: Optical Circuits Sharing for Software-Defined Data-Centers
- Distributed Dispatching in the Parallel Server Model
- DRIVE: One-bit Distributed Mean Estimation
- EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning
- Load Balancing with JET: Just Enough Tracking for Connection Consistency
- How to Send a Real Number Using a Single Bit (and Some Shared Randomness)
- Stochastic Coordination in Heterogeneous Load Balancing Systems
- SALSA: Self-Adjusting Lean Streaming Analytics
- Characterizing, Exploiting, and Detecting DMA Code Injection Vulnerabilities in the Presence of an IOMMU
- AnchorHash: A Scalable Consistent Hash
- Efficient Multiclass Classification with Duet