Elamur Research⌕
☰
home / papers / 2020 / Continual Relation Learning via Episodic Memory Activation and Reconsolidation

Continual Relation Learning via Episodic Memory Activation and Reconsolidation

Paper · 2020 · Xu Han, Yi Dai, Tianyu Gao et al.

Summary

Relation extraction models are usually trained once on a fixed set of relation types, but new relations keep appearing in real text. Continual relation learning addresses this by training a classifier on a sequence of tasks, each adding its own relations, while still requiring correct recognition of all earlier ones. Existing memory-based methods store a few past examples and replay them. However, they tend to overfit those stored instances and gradually blur distinctions between old relations.

The authors draw on the neuroscience finding that recalled memories must be reconsolidated to remain stable, and they propose EMAR. After the encoder is fine-tuned on a new task, the model is activated by replaying memorized and new examples together. A reconsolidation step then trains it to match instances to relation prototypes computed from memory. This constraint is looser than freezing old embeddings.

On FewRel, TACRED and SimpleQ, EMAR outperforms strong baselines, including the previous state of the art, EA-EMR, and it makes better use of limited memory. This helps keep extractors current without retraining on all historical data.

Key points

  • Evaluation protocol: After the k-th task, the model is tested on the union of all query sets seen so far and must classify among every known relation. The task therefore becomes harder as tasks accumulate.
  • Datasets and settings: Experiments use three benchmarks (FewRel, TACRED, SimpleQ) under two settings.
  • Baselines: EMR, EA-EMR, a consolidation-based method, and lower and upper bounds.
  • Main results: EMAR wins by a large margin on FewRel and TACRED. On SimpleQ it is only close to EMR and EA-EMR, since even the lower bound scores high there. A sizeable gap to the upper bound remains.
  • Memory size: Tested with 10, 25 and 50 stored examples per task. All models improve with more memory, but EMAR stays best at every size on FewRel and TACRED. It sometimes matches baselines that use larger memories.
  • Case study: The authors visualized two relations with 40 test examples each and measured separability with linear SVM boundaries. EA-EMR's L2 constraint packs features more densely over time, while EMAR's prototype-based reconsolidation keeps the relations more separable.
  • Code and data: Released at github.com/thunlp/ContinualRE.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Tsinghua UniversityBeijing · China137 works · 83,262 citations🏢TencentShenzhen · China66 works · 17,945 citations🔬State Key Laboratory of Intelligent Technology and SystemsBeijing · China21 works · 9,641 citations🔬Viện Trí tuệ nhân tạoHanoi · Vietnam2 works · 532 citations

📚 Shelves 4 Computer Science · 9,086 works Artificial Intelligence · 8,889 works Arts and Humanities · 9,296 works Language and Linguistics · 9,206 works

Cite

Cite as Xu Han, Yi Dai, Tianyu Gao, Yankai Lin, Zhiyuan Liu, Peng Li, Maosong Sun and Jie Zhou. 2020. Continual Relation Learning via Episodic Memory Activation and Reconsolidation. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2020.acl-main.573

@inproceedings{han2020continual,
  title = {{Continual Relation Learning via Episodic Memory Activation and Reconsolidation}},
  author = {Xu Han, Yi Dai, Tianyu Gao, Yankai Lin, Zhiyuan Liu, Peng Li, Maosong Sun and Jie Zhou},
  booktitle = {Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics},
  year = {2020},
  doi = {10.18653/v1/2020.acl-main.573},
  url = {https://doi.org/10.18653/v1/2020.acl-main.573}
}

References

78 references, 18 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Cristina M Alberini. 2005. Mechanisms of memory stabilization: are consolidation and reconsolidation similar or distinct processes? Trends in neurosciences, 28(1):51–56. find
  2. Rahaf Aljundi, Francesca Babiloni, Mohamed Elhoseiny, Marcus Rohrbach, and Tinne Tuytelaars. 2018. Memory aware synapses: Learning what (not) to forget. In Proceedings of ECCV, pages 139–154. find
  3. Gabor Angeli, Melvin Jose Johnson Premkumar, and Christopher D Manning. 2015. Leveraging linguistic structure for open domain information extraction. In Proceedings of ACL-IJCNLP, pages 344–354. our card
  4. Livio Baldini Soares, Nicholas FitzGerald, Jeffrey Ling, and Tom Kwiatkowski. 2019. Matching the blanks: Distributional similarity for relation learning. In Proceedings of ACL, pages 2895–2905. our card
  5. Michele Banko, Michael J Cafarella, Stephen Soderland, Matthew Broadhead, and Oren Etzioni. 2007. Open information extraction from the web. In Proceedings of IJCAI, pages 2670–2676. our card
  6. Hady Elsahar, Elena Demidova, Simon Gottschalk, Christophe Gravier, and Frederique Laforest. 2017. Unsupervised open relation extraction. In Proceedings of ESWC, pages 12–16. find
  7. Yoshua Bengio, Patrice Simard, and Paolo Frasconi. 1994. Learning long-term dependencies with gradient descent is difficult. IEEE transactions on neural networks, 5(2):157–166. find
  8. Oren Etzioni, Michael Cafarella, Doug Downey, Ana-Maria Popescu, Tal Shaked, Stephen Soderland, Daniel S Weld, and Alexander Yates. 2005. Unsupervised named-entity extraction from the web: An experimental study. AI, 165:91–134. find
  9. Bruno Bontempi, Catherine Laurent-Demir, Claude Destrade, and Robert Jaffard. 1999. Time-dependent reorganization of brain circuitry underlying long-term memory storage. Nature, 400(6745):671. find
  10. Anthony Fader, Stephen Soderland, and Oren Etzioni. 2011. Identifying relations for open information extraction. In Proceedings of EMNLP, pages 1535– 1545. find
  11. Antoine Bordes, Nicolas Usunier, Sumit Chopra, and Jason Weston. 2015. Large-scale simple question answering with memory networks. arXiv preprint arXiv:1506.02075. find
  12. Chrisantha Fernando, Dylan Banarse, Charles Blundell, Yori Zwols, David Ha, Andrei A Rusu, Alexander Pritzel, and Daan Wierstra. 2017. PathNet: Evolution channels gradient descent in super neural networks. ai.google. find
  13. Richard Boyce, Stephen D Glasgow, Sylvain Williams, and Antoine Adamantidis. 2016. Causal evidence for the role of rem sleep theta rhythm in contextual memory consolidation. Science, 352(6287):812– 816. find
  14. Razvan Bunescu and Raymond Mooney. 2007. Learning to extract relations from the web using minimal supervision. In Proceedings of ACL, pages 576– 583. our card
  15. Arslan Chaudhry, Marc’Aurelio Ranzato, Marcus Rohrbach, and Mohamed Elhoseiny. 2019. Efficient lifelong learning with a-gem. In Proceedings of ICLR. find
  16. Tianqi Chen, Ian Goodfellow, and Jonathon Shlens. 2016. Net2Net: Accelerating learning via knowledge transfer. In Proceedings of ICLR. find
  17. Lei Cui, Furu Wei, and Ming Zhou. 2018. Neural open information extraction. In Proceedings of ACL, pages 407–413. our card
  18. Cyprien de Masson d’Autume, Sebastian Ruder, Lingpeng Kong, and Dani Yogatama. 2019. Episodic memory in lifelong language learning. In Proceedings of NIPS. find
  19. Luciano Del Corro and Rainer Gemulla. 2013. Clausie: Clause-based open information extraction. In Proceedings of WWW, pages 355–366. find
  20. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of NAACL-HLT, pages 4171–4186. our card
  21. Li Dong, Furu Wei, Ming Zhou, and Ke Xu. 2015. Question answering over Freebase with multi-column convolutional neural networks. In Proceedings of ACL-IJCNLP, pages 260–269. find
  22. Tianyu Gao, Xu Han, Ruobing Xie, Zhiyuan Liu, Fen Lin, Leyu Lin, and Maosong Sun. 2020. Neural snowball for few-shot relation learning. In Proceedings of AAAI. find
  23. Matthew R Gormley, Mo Yu, and Mark Dredze. 2015. Improved relation extraction with feature-rich compositional embedding models. In Proceedings of EMNLP, pages 1774–1784. find
  24. Xu Han, Pengfei Yu, Zhiyuan Liu, Maosong Sun, and Peng Li. 2018a. Hierarchical relation extraction with coarse-to-fine grained attention. In Proceedings of EMNLP, pages 2236–2245. find
  25. Xu Han, Hao Zhu, Pengfei Yu, Ziyun Wang, Yuan Yao, Zhiyuan Liu, and Maosong Sun. 2018b. FewRel: A large-scale supervised few-shot relation classification dataset with state-of-the-art evaluation. In Proceedings of EMNLP, pages 4803–4809. find
  26. Raphael Hoffmann, Congle Zhang, Xiao Ling, Luke Zettlemoyer, and Daniel S Weld. 2011. Knowledge-based weak supervision for information extraction of overlapping relations. In Proceedings of ACL, pages 541–550. find
  27. Ronald Kemker and Christopher Kanan. 2018. Fearnet: Brain-inspired model for incremental learning. In Proceedings of ICLR. find
  28. James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska-Barwinska, et al. 2017. Overcoming catastrophic forgetting in neural networks. Proceedings of NAS, pages 3521–3526. find
  29. Jonathan LC Lee, Barry J Everitt, and Kerrie L Thomas. 2004. Independent cellular processes for hippocampal memory consolidation and reconsolidation. Science, 304(5672):839–843. find
  30. Zhizhong Li and Derek Hoiem. 2017. Learning without forgetting. TPAMI, 40(12):2935–2947. find
  31. Yankai Lin, Shiqi Shen, Zhiyuan Liu, Huanbo Luan, and Maosong Sun. 2016. Neural relation extraction with selective attention over instances. In Proceedings of ACL, pages 2124–2133. find
  32. ChunYang Liu, WenBo Sun, WenHan Chao, and Wanxiang Che. 2013. Convolution neural network for relation extraction. In Proceedings of ICDM, pages 231–242. find
  33. Thien Huu Nguyen and Ralph Grishman. 2015. Relation extraction: Perspective from convolutional neural networks. In Proceedings of the NAACL Workshop on Vector Space Modeling for NLP, pages 39– 48. our card
  34. Patrick Pantel and Marco Pennacchiotti. 2006. Espresso: Leveraging generic patterns for automatically harvesting semantic relations. In Proceedings of COLING, pages 113–120. our card
  35. Xialei Liu, Marc Masana, Luis Herranz, Joost Van de Weijer, Antonio M Lopez, and Andrew D Bagdanov. 2018. Rotate your networks: Better weight consolidation and less catastrophic forgetting. In Proceedings of ICPR, pages 2262–2268. find
  36. German I Parisi, Ronald Kemker, Jose L Part, Christopher Kanan, and Stefan Wermter. 2019. Continual lifelong learning with neural networks: A review. Neural Networks. our card
  37. Yang Liu, Furu Wei, Sujian Li, Heng Ji, Ming Zhou, and WANG Houfeng. 2015. A dependency-based neural network for relation classification. In Proceedings of ACL-IJCNLP, pages 285–290. find
  38. Fabio Petroni, Luciano Del Corro, and Rainer Gemulla. 2015. CORE: Context-aware open relation extraction with factorization machines. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1763–1773. find
  39. David Lopez-Paz and Marc’Aurelio Ranzato. 2017. Gradient episodic memory for continual learning. In Proceedings of NIPS, pages 6467–6476. find
  40. Diego Marcheggiani and Ivan Titov. 2016. Discrete-state variational autoencoders for joint discovery and factorization of relations. TACL, 4:231–244. find
  41. Mausam, Michael Schmitz, Stephen Soderland, Robert Bart, and Oren Etzioni. 2012. Open language learning for information extraction. In Proceedings of EMNLP-CoNLL, pages 523–534. find
  42. Mausam Mausam. 2016. Open information extraction systems and downstream applications. In Proceedings of IJCAI, pages 4074–4077. find
  43. James L McClelland, Bruce L McNaughton, and Randall C O’Reilly. 1995. Why there are complementary learning systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist models of learning and memory. Psychological review, 102(3):419. find
  44. Mike Mintz, Steven Bills, Rion Snow, and Dan Jurafsky. 2009. Distant supervision for relation extraction without labeled data. In Proceedings of ACL-IJCNLP, pages 1003–1011. find
  45. Makoto Miwa and Mohit Bansal. 2016. End-to-end relation extraction using LSTMs on sequences and tree structures. In Proceedings of ACL, pages 1105– 1116. our card
  46. Sylvestre-Alvise Rebuffi, Alexander Kolesnikov, Georg Sperl, and Christoph H Lampert. 2017. iCaRL: Incremental classifier and representation learning. In Proceedings of CVPR, pages 2001–2010. find
  47. Sebastian Riedel, Limin Yao, and Andrew McCallum. 2010. Modeling relations and their mentions without labeled text. In Proceedings of ECML-PKDD, pages 148–163. find
  48. Sebastian Riedel, Limin Yao, Andrew McCallum, and Benjamin M Marlin. 2013. Relation extraction with matrix factorization and universal schemas. In Proceedings of NAACL, pages 74–84. find
  49. Ellen Riloff, Rosie Jones, et al. 1999. Learning dictionaries for information extraction by multi-level bootstrapping. In Proceedings of AAAI, pages 474– 479. find
  50. Mark Bishop Ring. 1994. Continual learning in reinforcement environments. Ph.D. thesis, University of Texas at Austin Austin, Texas 78712. find
  51. Hippolyt Ritter, Aleksandar Botev, and David Barber. 2018. Online structured laplace approximations for overcoming catastrophic forgetting. In Proceedings of NIPS, pages 3738–3748. find
  52. Benjamin Rozenfeld and Ronen Feldman. 2008. Self-supervised relation extraction from the web. Proceedings of KAIS, 17(1):17–33. find
  53. Karim Nader, Glenn E Schafe, and Joseph E Le Doux. 2000. Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature, 406(6797):722. find
  54. Andrei A Rusu, Neil C Rabinowitz, Guillaume Desjardins, Hubert Soyer, James Kirkpatrick, Koray Kavukcuoglu, Razvan Pascanu, and Raia Hadsell. 2016. Progressive neural networks. ai.google. find
  55. Ndapandula Nakashole, Martin Theobald, and Gerhard Weikum. 2011. Scalable knowledge harvesting with high precision and high recall. In Proceedings of WSDM, pages 227–236. find
  56. Cicero dos Santos, Bing Xiang, and Bowen Zhou. 2015. Classifying relations by ranking with convolutional neural networks. In Proceedings of ACL-IJCNLP, pages 626–634. our card
  57. Michael Schlichtkrull, Thomas N Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, and Max Welling. 2018. Modeling relational data with graph convolutional networks. In Proceedings of ESWC, pages 593–607. find
  58. Limin Yao, Aria Haghighi, Sebastian Riedel, and Andrew McCallum. 2011. Structured relation discovery using generative models. In Proceedings of EMNLP, pages 1456–1466. our card
  59. Hanul Shin, Jung Kwon Lee, Jaehong Kim, and Jiwon Kim. 2017. Continual learning with deep generative replay. In Proceedings of NIPS, pages 2990–2999. find
  60. Mo Yu, Wenpeng Yin, Kazi Saidul Hasan, Cicero dos Santos, Bing Xiang, and Bowen Zhou. 2017. Improved neural relation detection for knowledge base question answering. In Proceedings of ACL, pages 571–581. find
  61. Yusuke Shinyama and Satoshi Sekine. 2006. Preemptive information extraction using unrestricted relation discovery. In Proceedings of NAACL-HLT, pages 304–311. our card
  62. Dmitry Zelenko, Chinatsu Aone, and Anthony Richardella. 2003. Kernel methods for relation extraction. Proceedings of JMLR, pages 1083–1106. our card
  63. Richard Socher, Brody Huval, Christopher D Manning, and Andrew Y Ng. 2012. Semantic compositionality through recursive matrix-vector spaces. In Proceedings of EMNLP, pages 1201–1211. our card
  64. Daojian Zeng, Kang Liu, Yubo Chen, and Jun Zhao. 2015. Distant supervision for relation extraction via piecewise convolutional neural networks. In Proceedings of EMNLP, pages 1753–1762. find
  65. Gabriel Stanovsky and Ido Dagan. 2016. Creating a large benchmark for open information extraction. In Proceedings of EMNLP, pages 2300–2305. find
  66. Daojian Zeng, Kang Liu, Siwei Lai, Guangyou Zhou, and Jun Zhao. 2014. Relation classification via convolutional deep neural network. In Proceedings of COLING, pages 2335–2344. our card
  67. Fan-Keng Sun, Cheng-Hao Ho, and Hung-Yi Lee. 2019. LAMAL: Language modeling is all you need for lifelong language learning. arXiv preprint arXiv:1909.03329. find
  68. Friedemann Zenke, Ben Poole, and Surya Ganguli. 2017. Continual learning through synaptic intelligence. In Proceedings of ICML, pages 3987–3995. find
  69. Sebastian Thrun and Lorien Pratt. 2012. Learning to learn. Springer Science & Business Media. find
  70. Yuhao Zhang, Victor Zhong, Danqi Chen, Gabor Angeli, and Christopher D Manning. 2017. Position-aware attention and supervised data improve slot filling. In Proceedings of EMNLP, pages 35–45. find
  71. Giulio Tononi and Chiara Cirelli. 2006. Sleep function and synaptic homeostasis. Sleep medicine reviews, 10(1):49–62. find
  72. Hong Wang, Wenhan Xiong, Mo Yu, Xiaoxiao Guo, Shiyu Chang, and William Yang Wang. 2019. Sentence embedding alignment for lifelong relation extraction. In Proceedings of NAACL-HLT, pages 796–806. find
  73. GuoDong Zhou, Jian Su, Jie Zhang, and Min Zhang. 2005. Exploring various knowledge in relation extraction. In Proceedings of ACL, pages 427–434. our card
  74. Jun Zhu, Zaiqing Nie, Xiaojiang Liu, Bo Zhang, and Ji-Rong Wen. 2009. StatSnowball: A statistical approach to extracting entity relationships. In Proceedings of WWW, pages 101–110. find
  75. Ruidong Wu, Yuan Yao, Xu Han, Ruobing Xie, Zhiyuan Liu, Fen Lin, Leyu Lin, and Maosong Sun. 2019. Open relation extraction: Relational knowledge transfer from supervised data to unsupervised data. In Proceedings of EMNLP-IJCNLP, pages 219–228. find
  76. Chenyan Xiong, Russell Power, and Jamie Callan. 2017. Explicit semantic ranking for academic search via knowledge graph embedding. In Proceedings of WWW, pages 1271–1279. find
  77. Yan Xu, Lili Mou, Ge Li, Yunchuan Chen, Hao Peng, and Zhi Jin. 2015. Classifying relations via long short term memory networks along shortest dependency paths. In Proceedings of EMNLP, pages 1785–1794. our card
  78. Guang Yang, Cora Sau Wan Lai, Joseph Cichon, Lei Ma, Wei Li, and Wen-Biao Gan. 2014. Sleep promotes branch-specific formation of dendritic spines after learning. Science, 344(6188):1173–1178. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/han-2020-continual-relation-learning-episodic-memory/