# To Softmax, or not to Softmax: that is the question when applying Active Learning for Transformer Models

arXiv: https://arxiv.org/abs/2210.03005 · HTML: https://mlsanity.com/paper/2210.03005/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 39 references that are in the map.
- First posted on arXiv 6 Oct 2022.

## Builds on (most cited first, 24 of 24)

- [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)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Rethinking the Inception Architecture for Computer Vision](https://mlsanity.com/paper/1512.00567/) (2015)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles](https://mlsanity.com/paper/1612.01474/) (2016)
- [HuggingFace's Transformers: State-of-the-art Natural Language Processing](https://mlsanity.com/paper/1910.03771/) (2019)
- [Evidential Deep Learning to Quantify Classification Uncertainty](https://mlsanity.com/paper/1806.01768/) (2018)
- [A Survey of Uncertainty in Deep Neural Networks](https://mlsanity.com/paper/2107.03342/) (2021)
- [Datasets: A Community Library for Natural Language Processing](https://mlsanity.com/paper/2109.02846/) (2021)
- [BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning](https://mlsanity.com/paper/1906.08158/) (2019)
- [Snorkel: Rapid Training Data Creation with Weak Supervision](https://mlsanity.com/paper/1711.10160/) (2017)
- [Why ReLU networks yield high-confidence predictions far away from the training data and how to mitigate the problem](https://mlsanity.com/paper/1812.05720/) (2018)
- [To Trust Or Not To Trust A Classifier](https://mlsanity.com/paper/1805.11783/) (2018)
- [Mix-n-Match: Ensemble and Compositional Methods for Uncertainty Calibration in Deep Learning](https://mlsanity.com/paper/2003.07329/) (2020)
- [Understanding Softmax Confidence and Uncertainty](https://mlsanity.com/paper/2106.04972/) (2021)
- [Mind Your Outliers! Investigating the Negative Impact of Outliers on Active Learning for Visual Question Answering](https://mlsanity.com/paper/2107.02331/) (2021)
- [Revisiting Uncertainty-based Query Strategies for Active Learning with Transformers](https://mlsanity.com/paper/2107.05687/) (2021)
- [Deep Learning using Rectified Linear Units (ReLU)](https://mlsanity.com/paper/1803.08375/) (2018)
- [Uncertainty Estimation for Language Reward Models](https://mlsanity.com/paper/2203.07472/) (2022)
- [Simple Techniques Work Surprisingly Well for Neural Network Test Prioritization and Active Learning (Replicability Study)](https://mlsanity.com/paper/2205.00664/) (2022)
- [Inhibited Softmax for Uncertainty Estimation in Neural Networks](https://mlsanity.com/paper/1810.01861/) (2018)
- [Small-Text: Active Learning for Text Classification in Python](https://mlsanity.com/paper/2107.10314/) (2021)
- [BayesFormer: Transformer with Uncertainty Estimation](https://mlsanity.com/paper/2206.00826/) (2022)

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

- [How to Fix a Broken Confidence Estimator: Evaluating Post-hoc Methods for Selective Classification with Deep Neural Networks](https://mlsanity.com/paper/2305.15508/) (2023)
- [ActiveGLAE: A Benchmark for Deep Active Learning with Transformers](https://mlsanity.com/paper/2306.10087/) (2023)
- [On Dataset Transferability in Active Learning for Transformers](https://mlsanity.com/paper/2305.09807/) (2023)

## Abstract (arXiv)

Despite achieving state-of-the-art results in nearly all Natural Language Processing applications, fine-tuning Transformer-based language models still requires a significant amount of labeled data to work. A well known technique to reduce the amount of human effort in acquiring a labeled dataset is textitActive Learning (AL): an iterative process in which only the minimal amount of samples is labeled. AL strategies require access to a quantified confidence measure of the model predictions. A common choice is the softmax activation function for the final layer. As the softmax function provides misleading probabilities, this paper compares eight alternatives on seven datasets. Our almost paradoxical finding is that most of the methods are too good at identifying the true most uncertain samples (outliers), and that labeling therefore exclusively outliers results in worse performance. As a heuristic we propose to systematically ignore samples, which results in improvements of various methods compared to the softmax function.