# Active Learning for Speech Recognition: the Power of Gradients

arXiv: https://arxiv.org/abs/1612.03226 · HTML: https://mlsanity.com/paper/1612.03226/

## Key facts

- Cited by 25 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 10 Dec 2016.

## Built on by (most cited first, 25 of 25)

- [Fine-Tuning Language Models from Human Preferences](https://mlsanity.com/paper/1909.08593/) (2019)
- [A Survey of Active Learning for Natural Language Processing](https://mlsanity.com/paper/2210.10109/) (2022)
- [Deep Active Learning by Leveraging Training Dynamics](https://mlsanity.com/paper/2110.08611/) (2021)
- [Diversity Measurement and Subset Selection for Instruction Tuning Datasets](https://mlsanity.com/paper/2402.02318/) (2024)
- [Unifying Approaches in Active Learning and Active Sampling via Fisher Information and Information-Theoretic Quantities](https://mlsanity.com/paper/2208.00549/) (2022)
- [Deep Active Learning with Noise Stability](https://mlsanity.com/paper/2205.13340/) (2022)
- [Multi-class Text Classification using BERT-based Active Learning](https://mlsanity.com/paper/2104.14289/) (2021)
- [Empowering Language Models with Active Inquiry for Deeper Understanding](https://mlsanity.com/paper/2402.03719/) (2024)
- [VL4Pose: Active Learning Through Out-Of-Distribution Detection For Pose Estimation](https://mlsanity.com/paper/2210.06028/) (2022)
- [A Neural Pre-Conditioning Active Learning Algorithm to Reduce Label Complexity](https://mlsanity.com/paper/2104.03525/) (2021)
- [Advancing Deep Active Learning & Data Subset Selection: Unifying Principles with Information-Theory Intuitions](https://mlsanity.com/paper/2401.04305/) (2024)
- [Distribution Aware Metrics for Conditional Natural Language Generation](https://mlsanity.com/paper/2209.07518/) (2022)
- [Investigating Multi-source Active Learning for Natural Language Inference](https://mlsanity.com/paper/2302.06976/) (2023)
- [Evaluating ASR Confidence Scores for Automated Error Detection in User-Assisted Correction Interfaces](https://mlsanity.com/paper/2503.15124/) (2025)
- [Deep Active Learning for Text Classification with Diverse Interpretations](https://mlsanity.com/paper/2108.10687/) (2021)
- [Smooth Sailing: Improving Active Learning for Pre-trained Language Models with Representation Smoothness Analysis](https://mlsanity.com/paper/2212.11680/) (2022)
- [Bayesian Uncertainty and Expected Gradient Length -- Regression: Two Sides Of The Same Coin?](https://mlsanity.com/paper/2104.09493/) (2021)
- [Active Learning for Natural Language Generation](https://mlsanity.com/paper/2305.15040/) (2023)
- [Active Learning to Classify Macromolecular Structures in situ for Less Supervision in Cryo-Electron Tomography](https://mlsanity.com/paper/2102.12040/) (2021)
- [Scalable Batch-Mode Deep Bayesian Active Learning via Equivalence Class Annealing](https://mlsanity.com/paper/2112.13737/) (2021)
- [ALLWAS: Active Learning on Language models in WASserstein space](https://mlsanity.com/paper/2109.01691/) (2021)
- [Nearest Neighbor Classifier with Margin Penalty for Active Learning](https://mlsanity.com/paper/2203.09174/) (2022)
- [Towards Efficient Active Learning in NLP via Pretrained Representations](https://mlsanity.com/paper/2402.15613/) (2024)
- [Combining X-Vectors and Bayesian Batch Active Learning: Two-Stage Active Learning Pipeline for Speech Recognition](https://mlsanity.com/paper/2406.02566/) (2024)
- [Geometric Data Valuation via Leverage Scores](https://mlsanity.com/paper/2511.02100/) (2025)

## Abstract (arXiv)

In training speech recognition systems, labeling audio clips can be expensive, and not all data is equally valuable. Active learning aims to label only the most informative samples to reduce cost. For speech recognition, confidence scores and other likelihood-based active learning methods have been shown to be effective. Gradient-based active learning methods, however, are still not well-understood. This work investigates the Expected Gradient Length (EGL) approach in active learning for end-to-end speech recognition. We justify EGL from a variance reduction perspective, and observe that EGL's measure of informativeness picks novel samples uncorrelated with confidence scores. Experimentally, we show that EGL can reduce word errors by 11%, or alternatively, reduce the number of samples to label by 50%, when compared to random sampling.