978 citations
- École Normale Supérieure - PSLFR306 papers
- Centre National de la Recherche ScientifiqueFR135 papers
- Institut national de recherche en sciences et technologies du numériqueFR94 papers
- Université Paris Sciences et LettresFR89 papers
- Université Paris CitéFR23 papers
- Sorbonne UniversitéFR19 papers
- SIERRA: Apprentissage Statistique et optimisationFR17 papers
- Université de MontréalCA15 papers
- École Polytechnique Fédérale de LausanneCH14 papers
- Czech Technical University in PragueCZ13 papers
- École PolytechniqueFR13 papers
- University of EdinburghGB13 papers
60 papers · 1 filter
In a Streaming World, Should You Stand Still? A Comprehensive Benchmark of Anomaly Detection in Streams
Magali Parrino, Antoine Ajenjo, Emmanuel Remy +3
Time series anomaly detection (TSAD) is increasingly deployed in streaming settings, where data arrive sequentially and may exhibit non-stationarity. As a result, several works fro…
Gaussian Flow-Matching Schedules: Implications for Sampling and Training
Arsène Claustre, Hugo Negrel, Claire Boyer +2
Flow-matching schedules affect both sampling dynamics and the variance of the regression target. For centered commuting Gaussians, we show that a direction-dependent schedule decom…
Unbiased Approximate Vector-Jacobian Products for Efficient Backpropagation
Killian Bakong, Laurent Massoulié, Edouard Oyallon +1
In this work we introduce methods to reduce the computational and memory costs of training deep neural networks. Our approach consists in replacing exact vector-jacobian products b…
Certified Per-Instance Unlearning Using Individual Sensitivity Bounds
Hanna Benarroch, Jamal Atif, Olivier Cappé
Certified machine unlearning can be achieved via noise injection leading to differential privacy guarantees, where noise is calibrated to worst-case sensitivity. Such conservative…
MSAD: A Deep Dive into Model Selection for Time series Anomaly Detection
Emmanouil Sylligardos, John Paparrizos, Themis Palpanas +2
Anomaly detection is a fundamental task for time series analytics with important implications for the downstream performance of many applications. Despite increasing academic inter…
Guided Flow Policy: Learning from High-Value Actions in Offline Reinforcement Learning
Franki Nguimatsia Tiofack, Théotime Le Hellard, Fabian Schramm +2
Offline reinforcement learning often relies on behavior regularization that enforces policies to remain close to the dataset distribution. However, such approaches fail to distingu…