Meta-Learning Approaches for Improving Detection of Unseen Speech Deepfakes
This addresses the need for robust detection systems against evolving speech deepfakes on social media, though it is incremental as it builds on existing meta-learning approaches.
The paper tackles the problem of speech deepfake detection generalizing to unseen attacks by using meta-learning to learn attack-invariant features, achieving an improvement in Equal Error Rate from 21.67% to 10.42% on the InTheWild dataset with only 96 samples.
Current speech deepfake detection approaches perform satisfactorily against known adversaries; however, generalization to unseen attacks remains an open challenge. The proliferation of speech deepfakes on social media underscores the need for systems that can generalize to unseen attacks not observed during training. We address this problem from the perspective of meta-learning, aiming to learn attack-invariant features to adapt to unseen attacks with very few samples available. This approach is promising since generating of a high-scale training dataset is often expensive or infeasible. Our experiments demonstrated an improvement in the Equal Error Rate (EER) from 21.67% to 10.42% on the InTheWild dataset, using just 96 samples from the unseen dataset. Continuous few-shot adaptation ensures that the system remains up-to-date.