# Selective Capability Unlearning in End-to-End Spoken Language Understanding

arXiv: https://arxiv.org/abs/2606.24063 · HTML: https://mlsanity.com/paper/2606.24063/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Machine unlearning, one of 12 directions in LLM safety and security.
- Builds on 21 of its 25 references that are in the map.
- First posted on arXiv 23 Jun 2026.

## Placed in

- LLM safety and security / [Machine unlearning](https://mlsanity.com/safety/machine-unlearning/)

## Builds on (most cited first, 21 of 21)

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [Extracting Training Data from Large Language Models](https://mlsanity.com/paper/2012.07805/) (2020)
- [Eternal Sunshine of the Spotless Net: Selective Forgetting in Deep Networks](https://mlsanity.com/paper/1911.04933/) (2019)
- [Amnesiac Machine Learning](https://mlsanity.com/paper/2010.10981/) (2020)
- [Large Language Model Unlearning](https://mlsanity.com/paper/2310.10683/) (2023)
- [Towards Unbounded Machine Unlearning](https://mlsanity.com/paper/2302.09880/) (2023)
- [SLURP: A Spoken Language Understanding Resource Package](https://mlsanity.com/paper/2011.13205/) (2020)
- [Machine Unlearning: Solutions and Challenges](https://mlsanity.com/paper/2308.07061/) (2023)
- [A Fine-tuned Wav2vec 2.0/HuBERT Benchmark For Speech Emotion Recognition, Speaker Verification and Spoken Language Understanding](https://mlsanity.com/paper/2111.02735/) (2021)
- [ESPnet-SLU: Advancing Spoken Language Understanding through ESPnet](https://mlsanity.com/paper/2111.14706/) (2021)
- [Recent Trends in Deep Learning Based Personality Detection](https://mlsanity.com/paper/1908.03628/) (2019)
- [Integration of Pre-trained Networks with Continuous Token Interface for End-to-End Spoken Language Understanding](https://mlsanity.com/paper/2104.07253/) (2021)
- [Speech-MASSIVE: A Multilingual Speech Dataset for SLU and Beyond](https://mlsanity.com/paper/2408.03900/) (2024)
- [Leveraging Acoustic and Linguistic Embeddings from Pretrained speech and language Models for Intent Classification](https://mlsanity.com/paper/2102.07370/) (2021)
- [SoK: Machine Unlearning for Large Language Models](https://mlsanity.com/paper/2506.09227/) (2025)
- ["Alexa, can you forget me?" Machine Unlearning Benchmark in Spoken Language Understanding](https://mlsanity.com/paper/2505.15700/) (2025)
- [Leveraging Pretrained ASR Encoders for Effective and Efficient End-to-End Speech Intent Classification and Slot Filling](https://mlsanity.com/paper/2307.07057/) (2023)
- [From Speech to Data: Unraveling Google's Use of Voice Data for User Profiling](https://mlsanity.com/paper/2403.05586/) (2024)

## Built on by (most cited first, 1 of 1)

- [Evaluating Machine Unlearning in ASR](https://mlsanity.com/paper/2609.34092/) (2026)

## Abstract (arXiv)

Modern spoken language understanding (SLU) systems are increasingly deployed in real-world settings, where specific functionalities may need to be removed due to policy or safety constraints. In SLU, a functionality corresponds to an intent and its associated slot-generation behavior. However, in autoregressive models, suppressing a target intent does not eliminate the conditional mapping that generates slots conditioned on that intent. When the intent prefix is externally supplied, the model can reconstruct the original intent-slot structure. We identify this structural failure as textbfcapability persistence. We propose textitunderlineBinding underlineSubspace (BSU), a representation-level framework that isolates and attenuates intent-conditioned directions underlying this mapping. Across SLU benchmarks, BSU substantially reduces forced-prefix recoverability while preserving retained performance.