mlsanityData to 14 Oct 2026

Mitigating Confounding in Speech-Based Dementia Detection through Weight Masking

arXiv 2506.05610 ↗First posted 5 Jun 2025cs.CLcited by 3

Not on the map. Why?

Abstract (arXiv)

Deep transformer models have been used to detect linguistic anomalies in patient transcripts for early Alzheimer's disease (AD) screening. While pre-trained neural language models (LMs) fine-tuned on AD transcripts perform well, little research has explored the effects of the gender of the speakers represented by these transcripts. This work addresses gender confounding in dementia detection and proposes two methods: the textitExtended Confounding Filter and the textitDual Filter, which isolate and ablate weights associated with gender. We evaluate these methods on dementia datasets with first-person narratives from patients with cognitive impairment and healthy controls. Our results show transformer models tend to overfit to training data distributions. Disrupting gender-related weights results in a deconfounded dementia classifier, with the trade-off of slightly reduced dementia detection performance.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 34 of its 54 references that are in the map. First posted on arXiv 5 Jun 2025.

Builds on

34 of 54 references.

References · 34

  1. 2017Attention Is All You Needcited by 60,361
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  3. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
  4. 2013Intriguing properties of neural networkscited by 5,155
  5. 2019DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lightercited by 4,307
  6. 2016Equality of Opportunity in Supervised Learningcited by 2,007
  7. 2019BioBERT: a pre-trained biomedical language representation model for biomedical text miningcited by 1,947
  8. 2016Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddingscited by 1,312
  9. 2019Learning Imbalanced Datasets with Label-Distribution-Aware Margin Losscited by 989
  10. 2018SNIP: Single-shot Network Pruning based on Connection Sensitivitycited by 812
  11. 2014Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Imagescited by 776
  12. 2024Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modificationscited by 231
22 more
  1. 2020Measuring and Reducing Gendered Correlations in Pre-trained Modelscited by 224
  2. 2020Social Biases in NLP Models as Barriers for Persons with Disabilitiescited by 182
  3. 2023Change is Hard: A Closer Look at Subpopulation Shiftcited by 120
  4. 2021Debiasing Pre-trained Contextualised Embeddingscited by 107
  5. 2020Alzheimer's Dementia Recognition through Spontaneous Speech: The ADReSS Challengecited by 104
  6. 2021FairFil: Contrastive Neural Debiasing Method for Pretrained Text Encoderscited by 75
  7. 2021Stereotype and Skew: Quantifying Gender Bias in Pre-trained and Fine-tuned Language Modelscited by 57
  8. 2021A Theory of Label Propagation for Subpopulation Shiftcited by 48
  9. 2020Achieving Equalized Odds by Resampling Sensitive Attributescited by 42
  10. 2021Your fairness may vary: Pretrained language model fairness in toxic text classificationcited by 39
  11. 2021A Novel Estimator of Mutual Information for Learning to Disentangle Textual Representationscited by 37
  12. 2021Contrastive Learning for Fair Representationscited by 34
  13. 2016Select-Additive Learning: Improving Generalization in Multimodal Sentiment Analysiscited by 29
  14. 2017On the Origin of Deep Learningcited by 23
  15. 2023Debiased Fine-Tuning for Vision-language Models by Prompt Regularizationcited by 21
  16. 2022Modular and On-demand Bias Mitigation with Attribute-Removal Subnetworkscited by 18
  17. 2021Disentangling Representations of Text by Masking Transformerscited by 15
  18. 2020A Tale of Two Perplexities: Sensitivity of Neural Language Models to Lexical Retrieval Deficits in Dementia of the Alzheimer's Typecited by 9
  19. 2024Effective Controllable Bias Mitigation for Classification and Retrieval using Gate Adapterscited by 8
  20. 2018Removing Confounding Factors Associated Weights in Deep Neural Networks Improves the Prediction Accuracy for Healthcare Applicationscited by 5
  21. 2023Backdoor Adjustment of Confounding by Provenance for Robust Text Classification of Multi-institutional Clinical Notescited by 5
  22. 2024Too Big to Fail: Larger Language Models are Disproportionately Resilient to Induction of Dementia-Related Linguistic Anomaliescited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026AutoSOTA: An End-to-End Automated Research System for State-of-the-Art AI Model Discoverycited by 23
  2. 2026When Prompts Interact: Assessing Prompt Arithmetic for Deconfounding under Distribution Shiftcited by 0
  3. 2026DeconDTN-Toolkit: A Library for Evaluation and Enhancement of Robustness to Provenance Shiftcited by 0