Learning from Noisy Labels for Entity-Centric Information Extraction

Abstract

Recent information extraction approaches have relied on training deep neural models. However, such models can easily overfit noisy labels and suffer from performance degradation. While it is very costly to filter noisy labels in large learning resources, recent studies show that such labels take more training steps to be memorized and are more frequently forgotten than clean labels, therefore are identifiable in training. Motivated by such properties, we propose a simple co-regularization framework for entity-centric information extraction, which consists of several neural models with identical structures but different parameter initialization. These models are jointly optimized with the task-specific losses and are regularized to generate similar predictions based on an agreement loss, which prevents overfitting on noisy labels. Extensive experiments on two widely used but noisy benchmarks for information extraction, TACRED and CoNLL03, demonstrate the effectiveness of our framework. We release our code to the community for future research 1 .

Improving Distantly Supervised Document-Level Relation Extraction Through Natural Language Inference

“…For future work, we plan to explore other strategies to better leverage the entity type information for RE with NLI and investigate if document-level NLI is also more challenging than sentence-level NLI. Another potential direction is to experiment with other DS techniques, such as integrating a denoising module to the RE model (Xiao et al, 2020) or using DS-trained models as a DS filter (Zhou and Chen, 2021).…”

New Frontiers of Information Extraction

“…We will introduce methodologies that enhance the robustness of learning systems for IE in both their learning and inference phases. Those methodologies involve self-supervised denoising techniques for training noise-robust IE models based on coregularized knowledge distillation (Zhou and Chen, 2021;Liang et al, 2021), label re-weighting (Wang et al, 2019b) and label smoothing (Lukasik et al, 2020). Besides, we will also discuss about unsuper-vised techniques for out-of-distribution (OOD) detection (Zhou et al, 2021b;Hendrycks et al, 2020), prediction with abstention (Dhamija et al, 2018;Hendrycks et al, 2018) and novelty class detection (Perera and Patel, 2019) that seek to help the IE model identify invalid inputs or inputs with semantic shifts during its inference phase.…”

OPI at SemEval-2022 Task 10: Transformer-based Sequence Tagging with Relation Classification for Structured Sentiment Analysis

“…Following the works on Named Entity Recognition problem (Akbik et al, 2018;Yamada et al, 2020;Zhou and Chen, 2021), we used micro-average F1 score as our main measure for the Extraction Module. In addition for this module we added a detailed measure for each tag type i.e.…”