Goto

Collaborating Authors

 Europe


Extracting Training Data from Molecular Pre-trained Models

Neural Information Processing Systems

This work, for the first time, explores the risks of extracting private training molecular data from molecular pre-trained models. This task is nontrivial as the molecular pre-trained models are non-generative and exhibit a diversity of model architectures, which differs significantly from language and image models.


Benchmarking the Attribution Quality of Vision Models Robin Hesse 1 Simone Schaub-Meyer 1,2 Stefan Roth 1,2 1 Department of Computer Science, Technical University of Darmstadt

Neural Information Processing Systems

Attribution maps are one of the most established tools to explain the functioning of computer vision models. They assign importance scores to input features, indicating how relevant each feature is for the prediction of a deep neural network.



He Did PR for Zuckerberg, Musk, and Google. Now He Says He 'Only Told Half the Story'

TIME - Tech

He Did PR for Zuckerberg, Musk, and Google. Now He Says He'Only Told Half the Story' Thirty thousand feet in the air, Mark Zuckerberg turned to his speechwriter. The duo were flying in Zuckerberg's jet to the United Nations General Assembly in New York, where the Facebook boss was scheduled to address world leaders. Zuckerberg had a question for his companion. "Wait, what exactly is the UN?" Dex Hunter-Torricke had to hide his surprise. Zuckerberg was, by this point in 2015, the head of a company that was reshaping politics and societies around the world, with 1.5 billion users and counting.


Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network

Neural Information Processing Systems

Real-world Image Dehazing (RID) aims to alleviate haze-induced degradation in real-world settings. This task remains challenging due to the complexities in accurately modeling real haze distributions and the scarcity of paired real-world data.