Temă de cercetare: Methods for Training Secure Artificial Intelligence Models

Improving Graph-Based Sleep Staging with Shared Waveform Node Representations

The study investigates the impact of node representation on the performance of graph models for sleep stage classification. The specific frontends of five architectures are replaced with the same one-dimensional convolutional network, trained directly on polysomnographic signals. Three training strategies are analyzed across five datasets…

Read MoreImproving Graph-Based Sleep Staging with Shared Waveform Node Representations

Towards Universal, Transferable, and Targeted Acoustic Attacks on ASR Systems

This paper proposes MuteOut, a method for generating universal adversarial perturbations that are applicable to multiple speech recordings and transferable across automatic speech recognition (ASR) models. The study analyzes the mechanisms through which common architectural features facilitate the transferability of these attacks, highlighting…

Read MoreTowards Universal, Transferable, and Targeted Acoustic Attacks on ASR Systems