Anne-Marie Kermarrec
Full Professor
Leads the Scalable Computing Systems lab, working on decentralized and federated learning, recommender systems and large-scale distributed systems.
Anne-Marie Kermarrec's lab at EPFL works on trustworthy & safe ML, signal processing & information theory and distributed, cloud & edge systems. 30 papers since 2023, 9 at top venues, mostly ICML and NeurIPS. 9 PhD students and 5 master's students. Applications close .
At a glance
- papers, 2023–2026
- 8/ year
- at T1 venues
- 30%
- PhD students
- 9
- MS students
- 5
- alumni listed
- —
Papers by yearfrom the lab homepage, 2023–2026
- 8 papers2026 · T1 3 · collecting
- 12 papers2025 · T1 2
- 5 papers2024 · T1 1
- 5 papers2023 · T1 3
T1 venuesOtherVenues: ICML, NeurIPS, AISTATS, ICLR, TKDE
Research focuspaper titles that use each topic's words
- decentralized10 papers
- Distributed, cloud & edge systems2 papers
- Trustworthy & safe ML1 paper
- Signal processing & information theory0 papers
- computing systems0 papers
Paper list
From the lab homepage.
20268 papers
- SRDSCommunication-Efficient Secure Aggregation in Decentralized Learning
- IPDPSHarMoEny: Efficient Inference of MoE Models (opens in a new tab)
- ICLR (T1)Navigating the Accuracy-Size Trade-Off with Flexible Model Merging
- EuroMLSys ’26: The Workshop on Machine Learning and SystemsOrbit: Efficient Agentic Inference using Priority Scheduling (opens in a new tab)
- ICML (T1)Position: Collaborative Agentic AI Needs Interoperability Across Ecosystems
- TMLR (T1)Practical and Private Hybrid ML Inference with Fully Homomorphic Encryption
- EuroMLSysRIVA: Leveraging LLM Agents for Reliable Configuration Drift Detection (opens in a new tab)
- EuroMLSys ’26: The Workshop on Machine Learning and SystemsThe Cost of Expertise: Understanding MoE Decode Performance (opens in a new tab)
202512 papers
- WWW (T1)Boosting Asynchronous Decentralized Learning with Model Fragmentation
- IEEE Transactions on Parallel and Distributed SystemsBoosting Resource-Constrained Federated Learning Systems with Guessed Updates (opens in a new tab)
- EuroMLSysEfficient Federated Search for Retrieval-Augmented Generation (opens in a new tab)
- International European Conference on Parallel and Distributed ComputingEfficient Pyramidal Analysis of Gigapixel Images on a Decentralized Modest Computer Cluster (opens in a new tab)
- IEEE Conference on Secure and Trustworthy Machine Learning (SaTML)Fair Decentralized Learning
- ALGOTEL - 27èmes Rencontres Francophones sur les Aspects Algorithmiques des TélécommunicationsFast & Fourier: tatouage spectral de graphes
- EuroMLSysHarnessing Increased Client Participation with Cohort-Parallel Federated Learning (opens in a new tab)
- International Middleware ConferenceLeveraging Approximate Caching for Faster Retrieval-Augmented Generation (opens in a new tab)
- The Privacy Enhancing Technologies SymposiumLow-Cost Privacy-Preserving Decentralized Learning (opens in a new tab)
- EuroMLSysPractical Federated Learning without a Server (opens in a new tab)
- SRDSRobust Fingerprinting of Graphs With Fing (opens in a new tab)
- ICML (T1)Robust ML Auditing using Prior Knowledge
20245 papers
- European Conference on Artificial IntelligenceFairness Auditing with Multi-Agent Collaboration (opens in a new tab)
- The Privacy Enhancing Technologies SymposiumNoiseless Privacy-Preserving Decentralized Learning (opens in a new tab)
- The International Symposium on Reliable Distributed SystemsPeerSwap: A Peer-Sampler with Randomness Guarantees
- ACM/IFIP International Middleware ConferenceQuickDrop: Efficient Federated Unlearning via Synthetic Data Generation
- NeurIPS (T1)Revisiting Ensembling in One-Shot Federated Learning
20235 papers
- Workshop on Machine Learning and SystemsDecentralized learning made easy with DecentralizePy (opens in a new tab)
- NeurIPS (T1)Epidemic Learning: Boosting Decentralized Learning with Randomized Communication
- ICDCS IEEE International Conference on Distributed Computing SystemsGet More for Less in Decentralized Learning Systems (opens in a new tab)
- TKDE (T1)GoldFinger: Fast & Approximate Jaccard for Efficient KNN Graph Constructions (opens in a new tab)
- AISTATS (T1)Refined Convergence and Topology Learning for Decentralized SGD with Heterogeneous Data
Where alumni went
Alumni are not listed on the lab homepage.
Apply to EPFLFall 2027
EDIC Computer and Communication Sciences PhD
- Round
- Round 1
- Opens
- —
- GRE
- optional
- English
- TOEFL 100 / IELTS 7.0
- Fee
- none
Confirmed on the program page.
Similar labsnearest by research summary
- USMichigan
Qiaozhu Mei
- Graph mining & networks
- USCMU
Carlee Joe-Wong
- Trustworthy & safe ML
- USStanford
Tsachy Weissman
- Learning theory & optimization
- USUIUC
Ning Luo
- Trustworthy & safe ML
- USUIUC
Hanghang Tong
- Graph mining & networks
- CNZhejiang
Lu Chen
- Graph mining & networks
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- The alumni list could not be found on the lab site.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
EPFLcloses