# DiDOTS: Knowledge Distillation from Large-Language-Models for Dementia Obfuscation in Transcribed Speech

arXiv: https://arxiv.org/abs/2410.04188 · HTML: https://mlsanity.com/paper/2410.04188/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge distillation for LLMs, one of 12 directions in Knowledge distillation.
- Builds on 39 of its 109 references that are in the map.
- First posted on arXiv 5 Oct 2024.

## Placed in

- Knowledge distillation / [Knowledge distillation for LLMs](https://mlsanity.com/distill/knowledge-distillation-for-llms/)

## Builds on (most cited first, 30 of 39)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Efficient Estimation of Word Representations in Vector Space](https://mlsanity.com/paper/1301.3781/) (2013)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://mlsanity.com/paper/2104.08691/) (2021)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [Enriching Word Vectors with Subword Information](https://mlsanity.com/paper/1607.04606/) (2016)
- [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://mlsanity.com/paper/1912.08777/) (2019)
- [Sequence-Level Knowledge Distillation](https://mlsanity.com/paper/1606.07947/) (2016)
- [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://mlsanity.com/paper/1909.05858/) (2019)
- [P-Tuning v2: Prompt Tuning Can Be Comparable to Fine-tuning Universally Across Scales and Tasks](https://mlsanity.com/paper/2110.07602/) (2021)
- [Adversarial Example Generation with Syntactically Controlled Paraphrase Networks](https://mlsanity.com/paper/1804.06059/) (2018)
- [Dear Sir or Madam, May I introduce the GYAFC Dataset: Corpus, Benchmarks and Metrics for Formality Style Transfer](https://mlsanity.com/paper/1803.06535/) (2018)
- [Style Transfer Through Back-Translation](https://mlsanity.com/paper/1804.09000/) (2018)
- [Utilizing BERT for Aspect-Based Sentiment Analysis via Constructing Auxiliary Sentence](https://mlsanity.com/paper/1903.09588/) (2019)
- [Alzheimer's Dementia Recognition through Spontaneous Speech: The ADReSS Challenge](https://mlsanity.com/paper/2004.06833/) (2020)
- [Detecting cognitive decline using speech only: The ADReSSo Challenge](https://mlsanity.com/paper/2104.09356/) (2021)
- [SMHD: A Large-Scale Resource for Exploring Online Language Usage for Multiple Mental Health Conditions](https://mlsanity.com/paper/1806.05258/) (2018)
- [Pre-trained Summarization Distillation](https://mlsanity.com/paper/2010.13002/) (2020)
- [Controlled Text Generation as Continuous Optimization with Multiple Constraints](https://mlsanity.com/paper/2108.01850/) (2021)
- [Mix and Match: Learning-free Controllable Text Generation using Energy Language Models](https://mlsanity.com/paper/2203.13299/) (2022)
- [A Dual Reinforcement Learning Framework for Unsupervised Text Style Transfer](https://mlsanity.com/paper/1905.10060/) (2019)
- [The Limits of Word Level Differential Privacy](https://mlsanity.com/paper/2205.02130/) (2022)
- [From LLM to NMT: Advancing Low-Resource Machine Translation with Claude](https://mlsanity.com/paper/2404.13813/) (2024)
- [A4NT: Author Attribute Anonymity by Adversarial Training of Neural Machine Translation](https://mlsanity.com/paper/1711.01921/) (2017)
- [Artificial Intelligence, speech and language processing approaches to monitoring Alzheimer's Disease: a systematic review](https://mlsanity.com/paper/2010.06047/) (2020)
- [A Systematic Study of Knowledge Distillation for Natural Language Generation with Pseudo-Target Training](https://mlsanity.com/paper/2305.02031/) (2023)

## Built on by (most cited first, 3 of 3)

- [Sentra-Guard: A Real-Time Multilingual Defense Against Adversarial LLM Prompts](https://mlsanity.com/paper/2510.22628/) (2025)
- [ClaritySpeech: Dementia Obfuscation in Speech](https://mlsanity.com/paper/2507.09282/) (2025)
- [Understanding Dementia Speech Alignment with Diffusion-Based Image Generation](https://mlsanity.com/paper/2508.09385/) (2025)

## Abstract (arXiv)

Dementia is a sensitive neurocognitive disorder affecting tens of millions of people worldwide and its cases are expected to triple by 2050. Alarmingly, recent advancements in dementia classification make it possible for adversaries to violate affected individuals' privacy and infer their sensitive condition from speech transcriptions. Existing obfuscation methods in text have never been applied for dementia and depend on the availability of large labeled datasets which are challenging to collect for sensitive medical attributes. In this work, we bridge this research gap and tackle the above issues by leveraging Large-Language-Models (LLMs) with diverse prompt designs (zero-shot, few-shot, and knowledge-based) to obfuscate dementia in speech transcripts. Our evaluation shows that LLMs are more effective dementia obfuscators compared to competing methods. However, they have billions of parameters which renders them hard to train, store and share, and they are also fragile suffering from hallucination, refusal and contradiction effects among others. To further mitigate these, we propose a novel method, DiDOTS. DiDOTS distills knowledge from LLMs using a teacher-student paradigm and parameter-efficient fine-tuning. DiDOTS has one order of magnitude fewer parameters compared to its teacher LLM and can be fine-tuned using three orders of magnitude less parameters compared to full fine-tuning. Our evaluation shows that compared to prior work DiDOTS retains the performance of LLMs achieving 1.3x and 2.2x improvement in privacy performance on two datasets, while humans rate it as better in preserving utility even when compared to state-of-the-art paraphrasing models.