Challenge: Methods addressing spurious correlations such as Just Train Twice involve reweighting a subset of the training set to maximize the worst-group accuracy.
Approach: They propose to reweight a subset of a training set to maximize the worst-group accuracy by detecting outliers and removing them before reweighing.
Outcome: The proposed method achieves competitive or better accuracy compared with JTT and can detect and remove annotation errors in the subset being reweighted in JTT.

Similar Papers

Towards Adversarially Robust Text Classifiers by Learning to Reweight Clean Examples (2022.findings-acl)

Copied to clipboard

Challenge: Existing defense methods improve the adversarial robustness by making models adapt to training set augmented with some adversarials.
Approach: They propose to introduce a reweighting mechanism to calibrate the training distribution to obtain robust models.
Outcome: The proposed method minimizes the loss of validation set mixed with clean examples and adversarial ones in an online learning manner.
On the Limitations of Dataset Balancing: The Lost Battle Against Spurious Correlations (2022.findings-naacl)

Copied to clipboard

Challenge: Recent work shows that deep learning models are sensitive to low-level correlations between simple features and specific output labels, leading to over-fitting and lack of generalization.
Approach: They propose to eliminate single-word correlations altogether to mitigate this problem . they highlight several alternatives to dataset balancing to enhance contexts .
Outcome: The proposed approach to balancing datasets is insufficient, the authors argue . they suggest enhancing datasets with richer contexts and abstaining from interaction .
Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasets (2022.acl-long)

Copied to clipboard

Challenge: Natural language processing models exploit spurious correlations between features and labels in datasets to perform well only within the distributions they are trained on.
Approach: They propose to generate a debiased version of a dataset and replace it with training data to train a model that is generalised to different task distributions.
Outcome: The proposed method outperforms or performs comparable to state-of-the-art debiasing strategies on a large suite of debiased, out-of distribution, and adversarial test sets.
Guide the Learner: Controlling Product of Experts Debiasing Method Based on Token Attribution Similarities (2023.eacl-main)

Copied to clipboard

Challenge: Several proposals have been put forward for improving out-of-distribution performance by mitigating dataset biases.
Approach: They propose a fine-tuning strategy that incorporates the similarity between the main and biased model attribution scores in a Product of Experts (PoE) loss function to further improve OOD performance.
Outcome: The proposed method improves OOD performance while maintaining in-distribution performance.
Unsupervised training data re-weighting for natural language understanding with local distribution approximation (2022.emnlp-industry)

Copied to clipboard

Challenge: a distribution mismatch between offline training and live data can cause biases . cyclic seasonality shifts, and changing pool of users can contribute to this problem .
Approach: They propose an unsupervised approach to mitigate offline training data sampling bias . they propose a local distribution approximation in the pre-trained embedding space .
Outcome: The proposed approach mitigates the offline training data sampling bias in multiple NLU tasks without additional annotation.
FIND: Human-in-the-Loop Debugging Deep Text Classifiers (2020.emnlp-main)

Copied to clipboard

Challenge: Existing models are limited in the number of available datasets and lack the necessary tools to improve them.
Approach: They propose a framework which enables humans to debug deep learning text classifiers by disabling irrelevant hidden features.
Outcome: Experiments show that using FIND, humans can improve CNN text classifiers trained on different types of imperfect datasets.
Identifying and Mitigating Spurious Correlations for Improving Robustness in NLP Models (2022.findings-naacl)

Copied to clipboard

Challenge: Existing work identifies task-specific shortcuts via human priors or error analyses, which requires extensive expertise and efforts.
Approach: They propose to automatically identify spurious correlations in NLP models at scale by using existing interpretability methods to extract tokens that significantly affect model’s decision process.
Outcome: The proposed method can identify spurious correlations in NLP models at scale and mitigate these leads to more robust models in multiple applications.
Relabeling Minimal Training Subset to Flip a Prediction (2024.findings-eacl)

Copied to clipboard

Challenge: Existing methods to identify and relabel training subsets that can flip a prediction are not efficient, argues a new study.
Approach: They propose an algorithm to identify and relabel the smallest training subset St needed to flip a prediction.
Outcome: The proposed algorithm can flip a prediction on a test point xt with 2% of training points . the proposed method can be used for multiple purposes including evaluating model robustness .
Fighting Bias With Bias: Promoting Model Robustness by Amplifying Dataset Biases (2023.findings-acl)

Copied to clipboard

Challenge: Recent work sought to develop robust, unbiased models by filtering biased examples from training sets.
Approach: They propose to filter out biased examples from training sets to improve models' performance.
Outcome: The proposed evaluation framework is more challenging than the original dataset splits and even more challenging that hand-crafted challenge sets.
Stubborn Lexical Bias in Data and Models (2023.findings-acl)

Copied to clipboard

Challenge: Recent work has focused on spurious correlations between features and labels in training data . but, we find strong evidence of corresponding bias in the trained models .
Approach: They propose a method to reduce spurious correlations in training data by reweighting it using a large pool of extracted features.
Outcome: The proposed method reduces spurious correlations in training data, but still finds strong evidence of bias in trained models.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations