mlsanityData to 14 Oct 2026

Rehearsal with Auxiliary-Informed Sampling for Audio Deepfake Detection

arXiv 2505.24486 ↗First posted 30 May 2025cs.SD cs.AI cs.CRcited by 0

Not on the map. Why?

Abstract (arXiv)

The performance of existing audio deepfake detection frameworks degrades when confronted with new deepfake attacks. Rehearsal-based continual learning (CL), which updates models using a limited set of old data samples, helps preserve prior knowledge while incorporating new information. However, existing rehearsal techniques don't effectively capture the diversity of audio characteristics, introducing bias and increasing the risk of forgetting. To address this challenge, we propose Rehearsal with Auxiliary-Informed Sampling (RAIS), a rehearsal-based CL approach for audio deepfake detection. RAIS employs a label generation network to produce auxiliary labels, guiding diverse sample selection for the memory buffer. Extensive experiments show RAIS outperforms state-of-the-art methods, achieving an average Equal Error Rate (EER) of 1.953 % across five experiences. The code is available at: https://github.com/falihgoz/RAIS.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 27 of its 32 references that are in the map. First posted on arXiv 30 May 2025.

Builds on

27 of 32 references.

References · 27

  1. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  2. 2020wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representationscited by 3,905
  3. 2016Learning without Forgettingcited by 2,496
  4. 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,333
  5. 2018Experience Replay for Continual Learningcited by 1,095
  6. 2023A Comprehensive Survey of Continual Learning: Theory, Method and Applicationcited by 872
  7. 2019Gradient based sample selection for online continual learningcited by 571
  8. 2019Online Continual Learning with Maximally Interfered Retrievalcited by 379
  9. 2021New Insights on Reducing Abrupt Representation Change in Online Continual Learningcited by 219
  10. 2021AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networkscited by 180
  11. 2018Continual Learning of Context-dependent Processing in Neural Networkscited by 179
  12. 2019ASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detectioncited by 178
15 more
  1. 2022Does Audio Deepfake Detection Generalize?cited by 135
  2. 2022Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentationcited by 117
  3. 2022Efficient Self-supervised Learning with Contextualized Target Representations for Vision, Speech and Languagecited by 110
  4. 2019Self-Supervised Generalisation with Meta Auxiliary Learningcited by 88
  5. 2023A Comprehensive Survey of Forgetting in Deep Learning Beyond Continual Learningcited by 69
  6. 2022Graph Lifelong Learning: A Surveycited by 30
  7. 2021Continual Learning for Fake Audio Detectioncited by 27
  8. 2020Light Convolutional Neural Network with Feature Genuinization for Detection of Synthetic Speech Attackscited by 23
  9. 2024Select and Distill: Selective Dual-Teacher Knowledge Transfer for Continual Learning on Vision-Language Modelscited by 19
  10. 2023Mitigating Task Interference in Multi-Task Learning via Explicit Task Routing with Non-Learnable Primitivescited by 18
  11. 2023What to Remember: Self-Adaptive Continual Learning for Audio Deepfake Detectioncited by 15
  12. 2022CFAD: A Chinese Dataset for Fake Audio Detectioncited by 14
  13. 2023Do You Remember? Overcoming Catastrophic Forgetting for Fake Audio Detectioncited by 11
  14. 2024Region-Based Optimization in Continual Learning for Audio Deepfake Detectioncited by 5
  15. 2024Freeze and Learn: Continual Learning with Selective Freezing for Speech Deepfake Detectioncited by 3

Built on by

Not cited yet.