technique

Integrated Gradients

techniqueactiveprovisionalintegrated-gradients-9ccab57b·1 events·first seen 7d ago

Aliases: Integrated Gradients

Co-occurring entities

CA-MHFA SLS AASIST WavLM What Do Deepfake Speech Detectors Actually Hear?ASVspoof 5

More like this (12)

gradient accumulation Evolved Policy Gradients Gradient Labs Gradient Reuse Textual Gradient Optimization conjugate gradient Policy Gradient Methods gradient noise scale gradient flow dynamics stochastic gradient ascent Bias Leaves a Gradient Trail gradient compression

Recent events (1)

5arXiv · cs.AI·7d ago·source ↗

Explainability pipeline reveals divergent cues used by deepfake speech detectors

Researchers propose an audio-native explainability pipeline using Integrated Gradients on time-aligned self-supervised representations to localize decision evidence in deepfake speech detectors. Applied to three WavLM-based detectors (AASIST, CA-MHFA, SLS) on the ASVspoof 5 benchmark, the method reveals that despite similar performance, each detector relies on fundamentally different cues: environmental noise, phoneme artifacts, and word boundaries respectively. Findings are validated via causal masking experiments that confirm performance degrades when primary cues are removed. The work advances interpretability of audio deepfake detection, relevant to AI safety and media authenticity.

Evaluation and Benchmarking AI Safety Research CA-MHFA Integrated Gradients SLS +4 more