--- id: qin26_interspeech title: "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 year: 2026 doi: 10.21437/Interspeech.2026-1042 isca_url: https://www.isca-archive.org/interspeech_2026/qin26_interspeech.html pdf_url: https://www.isca-archive.org/interspeech_2026/qin26_interspeech.pdf session: Speech Deepfake Detection, Attribution and Characterization topics: - speech-deepfake-detection - self-supervised - speaker-verification category: deepfake-security labels: - robustness-noise institutions: - Hong Kong Polytechnic University - Nanyang Technological University funding: - Innovation and Technology Fund of the Hong Kong SAR - National Key R&D Program of China code: url: "" license: "" open_to_collaboration: false wiki_frontmatter: id: qin26_interspeech category: deepfake-security labels: - robustness-noise institutions: - Hong Kong Polytechnic University - Nanyang Technological University updated: 2026-09-29 confidence: full-paper digest: v2 source: https://doi.org/10.21437/Interspeech.2026-1042 pdf: https://www.isca-archive.org/interspeech_2026/qin26_interspeech.pdf wiki_url: https://interspeech-2026-wiki.vercel.app/papers/qin26_interspeech/ markdown_url: https://interspeech-2026-wiki.vercel.app/papers/qin26_interspeech/markdown.md --- # 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* [PDF](https://www.isca-archive.org/interspeech_2026/qin26_interspeech.pdf) · [ISCA page](https://www.isca-archive.org/interspeech_2026/qin26_interspeech.html) · [DOI](https://doi.org/10.21437/Interspeech.2026-1042) **Category:** `deepfake-security` · **Labels:** `robustness-noise` **TL;DR** — This paper proposes Domain Gradient Surgery Meta-Learning (DGS-MLDG) and its layer-wise variant (LW-DGS-MLDG) to resolve gradient conflicts between meta-train and meta-test objectives in speech deepfake detection, achieving an average relative EER reduction of 5.29% over strong ERM baselines. ## Key contributions - Identified and analyzed gradient conflict between meta-train and meta-test objectives as a primary performance bottleneck in MLDG-based speech deepfake detectors. - Proposed Domain Gradient Surgery (DGS), an asymmetric projection strategy that removes destructive components from the meta-test gradient while anchoring to the meta-train gradient. - Developed Layer-Wise DGS (LW-DGS), an efficient variant that monitors cosine similarity and dynamically applies surgery only to conflict-prone layers. - Demonstrated consistent out-of-domain generalization improvements over ERM, standard MLDG, and symmetric multi-task gradient surgery baselines (PCGrad, GradVac, CAGrad). ## Problem Speech deepfake detectors experience catastrophic performance degradation in real-world deployments due to domain shifts like unseen synthesis attacks, codec artifacts, and transmission noise. Standard empirical risk minimization (ERM) pooling strategies fail to model domain structures explicitly. Although meta-learning domain generalization (MLDG) simulates domain shifts via meta-train and meta-test splits, naive gradient aggregation (NGA) causes destructive gradient cancellation because the opposing objectives frequently point in opposite directions. Symmetric multi-task methods like PCGrad are inadequate for bi-level meta-learning optimization, making principled directional projection necessary. ## Method The architecture combines an XLSR self-supervised learning (SSL) frontend, a linear projection layer, and a Bidirectional Mamba (BiMamba) classification head. MLDG splits source domains into meta-train (inner adaptation, loss L_mtr) and meta-test (generalization evaluation, loss L_mte) partitions. During training, DGS computes the cosine similarity between meta-train gradients (F = nabla L_mtr) and meta-test gradients (G = nabla L_mte). If the inner product is negative (langle G, F rangle < 0), DGS treats F as an invariant anchor and projects G onto the normal plane of F: G_proj = G - ((langle G, F rangle) / (||F||^2 + epsilon)) * F. This ensures a conflict-free optimization trajectory (langle G_proj, F rangle >= 0) while keeping the domain-specific meta-train gradient intact. Because global surgery is computationally heavy for 300M+ parameter backbones, LW-DGS performs independent, layer-wise conflict detection and dynamic surgery per parameter tensor theta_k using layer-level cosine similarity. Training uses the Adam optimizer with an outer learning rate of 10^-6, inner learning rate of 10^-2, batch size of 12, weight decay of 10^-4, and meta-test weighting beta = 0.5. Audio is cropped/concatenated to 4 seconds (64,600 samples) with RawBoost data augmentation. ## Experimental setup Trained on three datasets: ASVspoof 2019 LA (6 attack types), ASVspoof 5 (8 attack types), and CFAD (8 attack types), creating 22 distinct source domains. Evaluated on in-dataset benchmarks (ASVspoof 2021 DF, ASVspoof 5 test) and cross-dataset benchmarks (ADD 2023 R1 and R2, In-the-wild, CodecFake, and CFAD unseen test). Compared against ERM, standard MLDG, PCGrad, GradVac, and CAGrad using Equal Error Rate (EER) and average relative EER reduction. ## Results DGS-MLDG achieves the lowest cross-dataset EERs across nearly all benchmarks, yielding an average relative EER reduction of 5.29% over the ERM baseline. On the challenging In-the-wild dataset, DGS-MLDG lowers EER from 6.71% (ERM) and 6.47% (MLDG) down to 5.23%, and on CodecFake from 7.66% down to 6.76%. The layer-wise LW-DGS-MLDG variant achieves a comparable average relative improvement of 4.04% while operating selectively on sensitive layers. Symmetric multi-task baselines like PCGrad (6.17% on In-the-wild, 7.69% on CodecFake) and CAGrad underperform because they treat meta-train and meta-test objectives symmetrically and disrupt adaptation signals. | Systems | ASV21 (%) | ADD2023 R1 (%) | In-the-wild (%) | CodecFake (%) | Avg. Rel. Imp. (%) | |---|---|---|---|---|---| | ERM | 0.92 | 19.16 | 6.71 | 7.66 | - | | MLDG | 1.05 | 18.14 | 6.47 | 7.24 | 2.53 | | + PCGrad | - | - | 6.17 | 7.69 | - | | + GradVac | - | - | 6.34 | 7.40 | - | | + CAGrad | - | - | 7.40 | 7.36 | - | | DGS-MLDG (ours) | 1.00 | 17.76 | 5.23 | 6.76 | 5.29 | | LW-DGS-MLDG (ours) | 1.02 | 17.88 | 6.15 | 7.27 | 4.04 | ## Limitations The framework inherits the computational overhead characteristic of bi-level meta-learning optimization loops. Evaluation is constrained to speech deepfake detection datasets without scaling tests to massive open-world conversational audio streams. The interaction between layer-wise dynamic thresholds and diverse model backbones beyond XLSR-BiMamba requires further empirical validation. ## Why read this Researchers working on domain generalization, multi-task optimization, or speech deepfake detection should read this paper to understand why symmetric gradient surgery fails in bi-level meta-learning and how an asymmetric anchor-based projection protects adaptation signals. ## Code None released (as of this page's `updated` date). If you are an author with a repo, please claim this entry — see CONTRIBUTING.md. ## Applications Robust automated speaker verification security wrappers and real-time deepfake speech filtering systems deployed against unseen codecs, synthesis algorithms, and in-the-wild acoustic environments. ## Institutions / 機構 Hong Kong Polytechnic University, Nanyang Technological University **Funding / 經費:** Innovation and Technology Fund of the Hong Kong SAR, National Key R&D Program of China ## Related - [Domain-Adaptive Dual-Gating Mixture of Experts for Generalizable Speech Deepfake Detection](qin26b_interspeech.md) — same problem · relatedness 2.8/3 - [Improving Generalization in Speech Deepfake Detection via Orthogonality-Constrained Common-Specific Feature Decorrelation](kim26l_interspeech.md) — same problem · relatedness 2.8/3 - [Diffusion Reconstruction towards Generalizable Audio Deepfake Detection](cheng26_interspeech.md) — same problem · relatedness 2.8/3 - [GradHarmony: A Gradient Alignment and Magnitude Normalization Strategy for Audio Deepfake Detection](kim26r_interspeech.md) — same problem · relatedness 2.7/3 - [QAMO: Quality-aware Multi-centroid One-class Learning For Speech Deepfake Detection](truong26_interspeech.md) — same problem · relatedness 2.6/3 All 950k paper pairs scored by TypeSafe Jev (`scripts/related/`); relatedness 0 = unrelated … 3 = directly comparable.