Results 11 to 20 of about 265,325 (316)
Revisiting Gradient Clipping: Stochastic bias and tight convergence guarantees [PDF]
International Conference on Machine Learning, 2023 Gradient clipping is a popular modification to standard (stochastic) gradient descent, at every iteration limiting the gradient norm to a certain value $c>0$.
Anastasia Koloskova +2 more
semanticscholar +1 more source
Automatic Clipping: Differentially Private Deep Learning Made Easier and Stronger [PDF]
Neural Information Processing Systems, 2022 Per-example gradient clipping is a key algorithmic step that enables practical differential private (DP) training for deep learning models. The choice of clipping threshold R, however, is vital for achieving high accuracy under DP.
Zhiqi Bu +3 more
semanticscholar +1 more source
Exploring the Limits of Differentially Private Deep Learning with Group-wise Clipping [PDF]
International Conference on Learning Representations, 2022 Differentially private deep learning has recently witnessed advances in computational efficiency and privacy-utility trade-off. We explore whether further improvements along the two axes are possible and provide affirmative answers leveraging two ...
Jiyan He +8 more
semanticscholar +1 more source
Optimal Clipping and Magnitude-aware Differentiation for Improved Quantization-aware Training [PDF]
International Conference on Machine Learning, 2022 Data clipping is crucial in reducing noise in quantization operations and improving the achievable accuracy of quantization-aware training (QAT). Current practices rely on heuristics to set clipping threshold scalars and cannot be shown to be optimal. We
Charbel Sakr +5 more
semanticscholar +1 more source
American Journal of Neuroradiology, 2023 Surgical clipping is more effective than endovascular treatment of unruptured intracranial aneurysms in terms of the frequency of the primary outcome of treatment failure. Results were mainly driven by angiographic results at 1 year.
M. Chagnon +49 more
semanticscholar +1 more source
United European Gastroenterology journal, 2022 Background Direct and indirect clipping treatments are used worldwide to treat colonic diverticular bleeding (CDB), but their effectiveness has not been examined in multicenter studies with more than 100 cases. Objective We sought to determine the short‐
T. Kishino +44 more
semanticscholar +1 more source
Scaling up Differentially Private Deep Learning with Fast Per-Example Gradient Clipping [PDF]
Proceedings on Privacy Enhancing Technologies, 2020 Recent work on Renyi Differential Privacy has shown the feasibility of applying differential privacy to deep learning tasks. Despite their promise, however, differentially private deep networks often lag far behind their non-private counterparts in ...
Jaewoo Lee, Daniel Kifer
semanticscholar +1 more source
Surgical Clipping Versus Endovascular Coiling in the Management of Intracranial Aneurysms
Cureus, 2021 Intracranial aneurysms are pathological dilatations of intracranial arteries and prevail in around 3.2% of the general population. The worst outcome of an aneurysm is its rupture. Its prevention and management can be accomplished by two broad modalities:
Rishab Belavadi +7 more
semanticscholar +1 more source
TCL: an ANN-to-SNN Conversion with Trainable Clipping Layers [PDF]
Design Automation Conference, 2020 Spiking-neural-networks (SNNs) are promising at edge devices since the event-driven operations of SNNs provides significantly lower power compared to analog-neural-networks (ANNs).
Nguyen-Dong Ho, I. Chang
semanticscholar +1 more source
Autoclip: Adaptive Gradient Clipping for Source Separation Networks [PDF]
International Workshop on Machine Learning for Signal Processing, 2020 Clipping the gradient is a known approach to improving gradient descent, but requires hand selection of a clipping threshold hyperparameter. We present AutoClip, a simple method for automatically and adaptively choosing a gradient clipping threshold ...
Prem Seetharaman +3 more
semanticscholar +1 more source