Progressive Layer-wise Supervision: Deep-to-Shallow Supervision Annealing for Efficient and Robust Speech Deepfake Detection

Speech foundation models pretrained via self-supervised learning provide powerful representations for deepfake detection, yet we find that standard fine-tuning exploits only the deepest transformer layers-leaving the majority of the network's capacity discriminatively inert. Layer-wise probing of XLS-R-128 reveals that the top two-thirds of the transformer stack contribute negligibly to anti-spoofing performance after standard fine-tuning, a phenomenon we term shallow-layer dormancy. We connect this observation to the representational redundancy of pretrained transformers: adjacent layers encode highly similar information, and fine-tuning without explicit intermediate supervision reinforces rather than breaks this redundancy. To address this, we propose Progressive Layer-wise Supervision (PLS), a training framework that attaches a shared linear classifier to every transformer layer and schedules layer-wise loss contributions via an exponentially annealed power weighting. Early training concentrates supervision on deep layers to preserve the pretrained representational hierarchy; later training progressively redistributes supervision to shallower layers, converting dormant earlier representations into discriminative ones. Competitive detection (<9.25% average EER) is achievable on XLS-R-128 at layer 7, which reduces the parameter count from 315M to 101M and exemplifies an early exit mechanism. Experiments across twelve benchmarks spanning in-domain, and out-of-domain deepfake scenarios show PLS achieves a new state-of-the-art with 239M parameters, outperforming prior methods with larger and more complex architectures.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-16
DOI
https://doi.org/10.48550/arxiv.2611.20630
Primary Topic
Speech Recognition and Synthesis
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Progressive Layer-wise Supervision: Deep-to-Shallow Supervision Annealing for Efficient and Robust Speech Deepfake Detection

Xuechen Liu, Hoan My Tran, Xin Wang
arXiv (Cornell University)
Speech Recognition and Synthesis
preprint

Progressive Layer-wise Supervision: Deep-to-Shallow Supervision Annealing for Efficient and Robust Speech Deepfake Detection

Xuechen Liu, Hoan My Tran, Xin Wang
preprint en

Abstract

Speech foundation models pretrained via self-supervised learning provide powerful representations for deepfake detection, yet we find that standard fine-tuning exploits only the deepest transformer layers-leaving the majority of the network's capacity discriminatively inert. Layer-wise probing of XLS-R-128 reveals that the top two-thirds of the transformer stack contribute negligibly to anti-spoofing performance after standard fine-tuning, a phenomenon we term shallow-layer dormancy. We connect this observation to the representational redundancy of pretrained transformers: adjacent layers encode highly similar information, and fine-tuning without explicit intermediate supervision reinforces rather than breaks this redundancy. To address this, we propose Progressive Layer-wise Supervision (PLS), a training framework that attaches a shared linear classifier to every transformer layer and schedules layer-wise loss contributions via an exponentially annealed power weighting. Early training concentrates supervision on deep layers to preserve the pretrained representational hierarchy; later training progressively redistributes supervision to shallower layers, converting dormant earlier representations into discriminative ones. Competitive detection (<9.25% average EER) is achievable on XLS-R-128 at layer 7, which reduces the parameter count from 315M to 101M and exemplifies an early exit mechanism. Experiments across twelve benchmarks spanning in-domain, and out-of-domain deepfake scenarios show PLS achieves a new state-of-the-art with 239M parameters, outperforming prior methods with larger and more complex architectures.

arXiv (Cornell University)
Speech Recognition and Synthesis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon