An Empirical Investigation of Learning from Biased Toxicity Labels

• URL: http://arxiv.org/abs/2110.01577v1
• Date: Mon, 4 Oct 2021 17:19:57 GMT
• Title: An Empirical Investigation of Learning from Biased Toxicity Labels
• Authors: Neel Nanda, Jonathan Uesato, Sven Gowal
• Abstract summary: We study how different training strategies can leverage a small dataset of human-annotated labels and a large but noisy dataset of synthetically generated labels. We evaluate the accuracy and fairness properties of these approaches, and trade-offs between the two.
• Score: 15.822714574671412
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Collecting annotations from human raters often results in a trade-off between the quantity of labels one wishes to gather and the quality of these labels. As such, it is often only possible to gather a small amount of high-quality labels. In this paper, we study how different training strategies can leverage a small dataset of human-annotated labels and a large but noisy dataset of synthetically generated labels (which exhibit bias against identity groups) for predicting toxicity of online comments. We evaluate the accuracy and fairness properties of these approaches, and trade-offs between the two. While we find that initial training on all of the data and fine-tuning on clean data produces models with the highest AUC, we find that no single strategy performs best across all fairness metrics.

Related papers

• Learning with Confidence: Training Better Classifiers from Soft Labels [0.0]
  In supervised machine learning, models are typically trained using data with hard labels, i.e., definite assignments of class membership. We investigate whether incorporating label uncertainty, represented as discrete probability distributions over the class labels, improves the predictive performance of classification models.
  arXiv  Detail & Related papers  (2024-09-24T13:12:29Z)
• Virtual Category Learning: A Semi-Supervised Learning Method for Dense Prediction with Extremely Limited Labels [63.16824565919966]
  This paper proposes to use confusing samples proactively without label correction. A Virtual Category (VC) is assigned to each confusing sample in such a way that it can safely contribute to the model optimisation. Our intriguing findings highlight the usage of VC learning in dense vision tasks.
  arXiv  Detail & Related papers  (2023-12-02T16:23:52Z)
• GaussianMLR: Learning Implicit Class Significance via Calibrated Multi-Label Ranking [0.0]
  We propose a novel multi-label ranking method: GaussianMLR. It aims to learn implicit class significance values that determine the positive label ranks. We show that our method is able to accurately learn a representation of the incorporated positive rank order.
  arXiv  Detail & Related papers  (2023-03-07T14:09:08Z)
• An Effective Approach for Multi-label Classification with Missing Labels [8.470008570115146]
  We propose a pseudo-label based approach to reduce the cost of annotation without bringing additional complexity to the classification networks. By designing a novel loss function, we are able to relax the requirement that each instance must contain at least one positive label. We show that our method can handle the imbalance between positive labels and negative labels, while still outperforming existing missing-label learning approaches.
  arXiv  Detail & Related papers  (2022-10-24T23:13:57Z)
• How many labelers do you have? A closer look at gold-standard labels [10.637125300701795]
  We show how access to non-aggregated label information can make training well-calibrated models more feasible than it is with gold-standard labels. We make several predictions for real-world datasets, including when non-aggregate labels should improve learning performance.
  arXiv  Detail & Related papers  (2022-06-24T02:33:50Z)
• One Positive Label is Sufficient: Single-Positive Multi-Label Learning with Label Enhancement [71.9401831465908]
  We investigate single-positive multi-label learning (SPMLL) where each example is annotated with only one relevant label. A novel method named proposed, i.e., Single-positive MultI-label learning with Label Enhancement, is proposed. Experiments on benchmark datasets validate the effectiveness of the proposed method.
  arXiv  Detail & Related papers  (2022-06-01T14:26:30Z)
• Acknowledging the Unknown for Multi-label Learning with Single Positive Labels [65.5889334964149]
  Traditionally, all unannotated labels are assumed as negative labels in single positive multi-label learning (SPML) We propose entropy-maximization (EM) loss to maximize the entropy of predicted probabilities for all unannotated labels. Considering the positive-negative label imbalance of unannotated labels, we propose asymmetric pseudo-labeling (APL) with asymmetric-tolerance strategies and a self-paced procedure to provide more precise supervision.
  arXiv  Detail & Related papers  (2022-03-30T11:43:59Z)
• Debiased Pseudo Labeling in Self-Training [77.83549261035277]
  Deep neural networks achieve remarkable performances on a wide range of tasks with the aid of large-scale labeled datasets. To mitigate the requirement for labeled data, self-training is widely used in both academia and industry by pseudo labeling on readily-available unlabeled data. We propose Debiased, in which the generation and utilization of pseudo labels are decoupled by two independent heads.
  arXiv  Detail & Related papers  (2022-02-15T02:14:33Z)
• Improving Contrastive Learning on Imbalanced Seed Data via Open-World Sampling [96.8742582581744]
  We present an open-world unlabeled data sampling framework called Model-Aware K-center (MAK) MAK follows three simple principles: tailness, proximity, and diversity. We demonstrate that MAK can consistently improve both the overall representation quality and the class balancedness of the learned features.
  arXiv  Detail & Related papers  (2021-11-01T15:09:41Z)
• Disentangling Sampling and Labeling Bias for Learning in Large-Output Spaces [64.23172847182109]
  We show that different negative sampling schemes implicitly trade-off performance on dominant versus rare labels. We provide a unified means to explicitly tackle both sampling bias, arising from working with a subset of all labels, and labeling bias, which is inherent to the data due to label imbalance.
  arXiv  Detail & Related papers  (2021-05-12T15:40:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.