On Robustness and Bias Analysis of BERT-based Relation Extraction
arXiv:2009.06206
Abstract
Fine-tuning pre-trained models have achieved impressive performance on standard natural language processing benchmarks. However, the resultant model generalizability remains poorly understood. We do not know, for example, how excellent performance can lead to the perfection of generalization models. In this study, we analyze a fine-tuned BERT model from different perspectives using relation extraction. We also characterize the differences in generalization techniques according to our proposed improvements. From empirical experimentation, we find that BERT suffers a bottleneck in terms of robustness by way of randomizations, adversarial and counterfactual tests, and biases (i.e., selection and semantic). These findings highlight opportunities for future improvements. Our open-sourced testbed DiagnoseRE is available in \url{https://github.com/zjunlp/DiagnoseRE}.
arXiv commentwork in progress
References in corpus (11)
- Axiomatic Attribution for Deep Networks
- A Closer Look at Memorization in Deep Networks
- Reinforcement Learning for Relation Classification from Noisy Data
- Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment
- HotFlip: White-Box Adversarial Examples for Text Classification
- Conceptualized Representation Learning for Chinese Biomedical Text Mining
- Stiffness: A New Perspective on Generalization in Neural Networks
- Extracting Multiple-Relations in One-Pass with Pre-Trained Transformers
- A Closer Look at Data Bias in Neural Extractive Summarization Models
- Probing Linguistic Features of Sentence-Level Representations in Neural Relation Extraction
- Bridging Text and Knowledge with Multi-Prototype Embedding for Few-Shot Relational Triple Extraction