Reinforcement Learning for Relation Classification From Noisy Data

By Jun Feng, Minlie Huang, Li Zhao, Yang Yang, Xiaoyan Zhu. AAAI Conference on Artificial Intelligence, 2018

353 citations recorded by the source corpus.

Existing relation classification methods that rely on distant supervision assume that a bag of sentences mentioning an entity pair are all describing a relation for the entity pair. Such methods, performing classification at the bag level, cannot identify the mapping between a relation and a sentence, and largely suffers from the noisy labeling problem. In this paper, we propose a novel model for relation classification at the sentence level from noisy data. The model has two modules: an instance selector and a relation classifier. The instance selector chooses high-quality sentences with reinforcement learning and feeds the selected sentences into the relation classifier, and the relation classifier makes sentence-level prediction and provides rewards to the instance selector. The two modules are trained jointly to optimize the instance selection and relation classification processes.Experiment results show that our model can deal with the noise of data effectively and obtains better performance for relation classification at the sentence level.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/b3dae9529f3caeeec9cc6872e94aa690418acb22

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Zhiyuan Liu
2. Maosong Sun
3. Xu Han
4. Mohit Bansal
5. Jie Zhou
6. Tianyu Gao
7. William Yang Wang
8. Jianfeng Gao

Citations in the corpus

• Hybrid Attention-Based Prototypical Networks for Noisy Few-Shot Relation Classification (2019, background): “Because DS brings inevitable noise in itself, many efforts are devoted to further reducing noise (Wu, Bamman, and Russell 2017; Feng et al. 2018).”
• A Survey on Knowledge Graphs: Representation, Acquisition, and Applications (2020, neutral): “[142] proposed different reward strategies.”
• More Data, More Relations, More Context and More Openness: A Review and Outlook for Relation Extraction (2020, methodology): “Furthermore, reinforcement learning (Feng et al., 2018; Zeng et al., 2018) and adver-”
• Joint Entity Linking with Deep Reinforcement Learning (2019, methodology): “By this property, RL has been successfully used in many NLP tasks, such as information retrieval [28], dialogue system [10] and relation classification [12], etc.”
• Hierarchical Relation Extraction with Coarse-to-Fine Grained Attention (2018, methodology): “More sophisticated mechanisms, such as reinforcement learning (Feng et al., 2018; Zeng et al., 2018) and adversarial training (Wu et al., 2017), have also been adapted for RE recently.”
• Dialog State Tracking with Reinforced Data Augmentation (2019, methodology): “Previous works by (Feng et al., 2018) and (Qin et al., 2018b) design RL algorithm to learn how to fil-ter out noisy ones.”
• Distant Supervision Relation Extraction with Intra-Bag and Inter-Bag Attentions (2019, neutral): “Feng et al. (2018) and Qin et al.”
• FewRel: A Large-Scale Supervised Few-Shot Relation Classification Dataset with State-of-the-Art Evaluation (2018, methodology): “…is followed by the recent state-of-the-art methods (Zeng et al., 2017; Ji et al., 2017; Huang and Wang, 2017; Wu et al., 2017; Liu et al., 2017; Feng et al., 2018; Zeng et al., 2018), on the benchmark dataset NYT-10 NYT-10 ( Riedel We formulate RC as a few-shot learning task in this paper,…”
• Reinforced Multi-Teacher Selection for Knowledge Distillation (2020, methodology): “Rand-Single-Ensemble Teacher uses the strategy of Fukuda et al. (2017) to randomly select one teacher from the teacher model candidates at mini-batch level to provide soft-targets for training student model.”
• Neural Learning of One-of-Many Solutions for Combinatorial Problems in Structured Output Spaces (2020, methodology): “…Learning’ (MIL) approach for Relation Extraction (RE) which employs a selection module to pick a set of sentences to be used for 2Available at https://data.dgl.ai/models/rrn-sudoku.pkl training a relation classifier, given a set of noisy relation labels [Feng et al., 2018, Qin et al., 2018].”
• Zero-shot Text Classification via Reinforced Self-training (2020, background): “(Feng et al., 2018; Qin et al., 2018) proposed a policy network to automatically identify wrongly-labeled instances in training set.”
• Towards Understanding Gender Bias in Relation Extraction (2019, methodology): “Great efforts have been made to improve NRE models by mitigating the effects of noise in the training data introduced by Distant Supervision (Hoffmann et al., 2011; Surdeanu et al., 2012; Lin et al., 2016; Liu et al., 2017; Feng et al., 2018; Qin et al., 2018).”
• Long-tail Relation Extraction via Knowledge Graph Embeddings and Graph Convolution Networks (2019, methodology): “We evaluate our models on the NYT dataset developed by (Riedel et al., 2010), which has been widely used in recent studies (Lin et al., 2016; Liu et al., 2017; Wu et al., 2017; Feng et al., 2018).”