DGS-MLDG: Domain Gradient Surgery Guided Meta-Learning for Domain Generalization in Speech Deepfake Detection

Siqing Qin; Kong Aik Lee; Youzhi Tu; Eng Siong Chng; Man-Wai Mak

eess.ASSep 27, 2026

DGS-MLDG: Domain Gradient Surgery Guided Meta-Learning for Domain Generalization in Speech Deepfake Detection

Authors: Siqing Qin, Kong Aik Lee, Youzhi Tu, Eng Siong Chng, Man-Wai Mak

Organizations: Dept. of Electrical and Electronic Engineering, The Hong Kong Polytechnic University · College of Computing and Data Science, Nanyang Technological University

Abstract

Speech deepfake detection faces significant challenges due to domain shifts. Domain generalization (DG), particularly meta-learning for domain generalization (MLDG), offers a promising solution by simulating and mitigating domain shifts. However, MLDG is often hindered by conflicting gradients between its meta-train and meta-test objectives, leading to suboptimal performance. To address this problem, we propose domain gradient surgery (DGS), a meta-learning method that resolves conflicts through an asymmetric projection strategy. DGS removes the destructive component from the meta-test gradient, ensuring a conflict-free optimization trajectory versus the meta-train gradient. Furthermore, we introduce layer-wise DGS (LW-DGS), an efficient variant of DGS that dynamically identifies and intervenes only conflict-prone layers. Extensive experiments on challenging benchmarks demonstrate that DGS-MLDG and LW-DGS-MLDG achieve an average relative EER reduction of 5.29% and 4.04%, respectively.

Figures & tables

Figure 1: Negative cosine similarity between the meta-train gradient and meta-test gradient of MLDG during training.

Figure 2: Overview of the proposed DGS-MLDG framework. (a) The training pipeline simulates domain shift via meta-splits and resolving conflicts via the selective mechanism (SM). (b) The backbone architecture used for deepfake detection. (c) Geometric comparison showing how DGS corrects the update direction by projecting conflicting gradients, avoiding the oscillation of Naive Gradient Aggregation (NGA).

Methods	ASV21 $\downarrow$	ASV5 $\downarrow$	CFAD $\downarrow$	ADD2023 R1 $\downarrow$	ADD2023 R2 $\downarrow$	In-the-wild $\downarrow$	CodecFake $\downarrow$	Avg. Rel. Imp. ( $\%$ ) $\uparrow$
ERM	0.92	8.57	24.33	19.16	23.23	6.71	7.66	-
MLDG	1.05	9.69	24.16	18.14	23.03	6.47	7.24	2.53
DGS-MLDG (ours)	1.00	10.24	23.86	17.76	23.19	5.23	6.76	5.29
LW-DGS-MLDG (ours)	1.02	9.55	23.83	17.88	22.68	6.15	7.27	4.04

Table 1: Comparison of EER (%) among main methods. The last column shows the cross-dataset average relative improvement w.r.t. the ERM baseline (

\uparrow

). Bold and underline denote the lowest and the second-lowest EERs for each dataset.

Experiments	ASV21 $\downarrow$	ASV5 $\downarrow$	CFAD $\downarrow$	ADD2023 R1 $\downarrow$	ADD2023 R2 $\downarrow$	In-the-wild $\downarrow$	CodecFake $\downarrow$
Ablation ①	0.97	9.82	24.27	20.43	23.68	6.15	7.27
Ablation ②	1.16	9.57	23.93	18.03	22.75	6.58	7.27

Table 2: Ablation study: EER (%) under all evaluation settings.

System	In-the-wild	CodecFake
XLSR-Mamba	6.71	7.66
w/ MLDG training	6.47	7.24
+ DGS (ours)	5.23	6.76
+ LW-DGS (ours)	6.15	7.27
+ PCGrad	6.17	7.69
+ GradVac	6.34	7.40

Table 3: Comparison of EER (%) results across different gradient alignment methods.

Figure 3: Cosine similarity comparison across MLDG, DGS-MLDG, and LW-DGS-MLDG. Conflicts decrease via proposed DGS and LW-DGS.

Explore similar work

Domain-Adaptive Dual-Gating Mixture of Experts for Generalizable Speech Deepfake Detection
Sep 27, 2026Siqing Qin, Zhe Li, Kong Aik Lee +1Domain-Adapted Diffusion Acoustic
Dept. of Electrical and Electronic Engineering, The Hong Kong Polytechnic University · Speech, Language, and Cognition Laboratory, The University of Hong Kong
GLAD: Global-Local Adaptive Detector for Robust Speech Deepfake Detection
Sep 28, 2026Zelin Zhao, Guanjie Huang, Danny Hin Kwok Tsang +1Audio Deepfake Detection Speech Synthesis
The Hong Kong University of Science and Technology (Guangzhou)
Mitigating Proxy-to-Wild Domain Gap in Deepfake Speech
Jun 5, 2026Xuanjun Chen, Yun-Shing Wu, Wei-Chung Lu +4
Graduate Institute of Communication Engineering, National Taiwan University · Graduate Institute of Networking and Multimedia, National Taiwan University · Department of Information Management, National Taiwan University +2

Sep 27, 2026, eess.ASJ/K move · Enter open · S save

Domain-Adaptive Dual-Gating Mixture of Experts for Generalizable Speech Deepfake Detection

Siqing Qin, Zhe Li, Kong Aik Lee, Man-Wai Mak

Dept. of Electrical and Electronic Engineering, The Hong Kong Polytechnic University · Speech, Language, and Cognition Laboratory, The University of Hong Kong

Domain-Adapted Diffusion Acoustic Experts

Open paper AskPDF arXiv

Recent advances in speech deepfake detection (SDD) have leveraged the Mixture of Experts (MoE) to enhance generalization capacity. However, existing gating networks often overlook the acoustic and temporal cues of deepfakes. In this work, we propose a novel domain-adaptive dual-gating MoE (DADGMoE) framework for SDD under unseen attack types and acoustic conditions. Our innovative dual-gating mechanism leverages Sinc-layer-based filters to process both low-level acoustic signals (raw waveforms) and high-level speech representations from a large self-supervised learning (SSL) model. It further incorporates domain prototypes to guide expert routing based on implicit deepfake patterns. The lightweight affine experts process the routed inputs. Experiments show that our DADGMoE significantly outperforms the baseline, achieving up to a 40.8% relative EER reduction on challenging out-of-dataset benchmarks. This framework demonstrates superior generalization capabilities and efficient design.

Abstract

Explore similar work

Domain-Adaptive Dual-Gating Mixture of Experts for Generalizable Speech Deepfake Detection

GLAD: Global-Local Adaptive Detector for Robust Speech Deepfake Detection

Mitigating Proxy-to-Wild Domain Gap in Deepfake Speech