Continual Relation Learning via Episodic Memory Activation and Reconsolidation

By Xu Han, Yi Dai, Tianyu Gao, Yankai Lin, Zhiyuan Liu, Peng Li, Maosong Sun, Jie Zhou. 2020

124 citations recorded by the source corpus.

Continual relation learning aims to continually train a model on new data to learn incessantly emerging novel relations while avoiding catastrophically forgetting old relations. Some pioneering work has proved that storing a handful of historical relation examples in episodic memory and replaying them in subsequent training is an effective solution for such a challenging problem. However, these memory-based methods usually suffer from overfitting the few memorized examples of old relations, which may gradually cause inevitable confusion among existing relations. Inspired by the mechanism in human long-term memory formation, we introduce episodic memory activation and reconsolidation (EMAR) to continual relation learning. Every time neural models are activated to learn both new and memorized data, EMAR utilizes relation prototypes for memory reconsolidation exercise to keep a stable understanding of old relations. The experimental results show that EMAR could get rid of catastrophically forgetting old relations and outperform the state-of-the-art continual learning models.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Meta & Continual Learning, Knowledge Graphs & Representation

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

In Computer Vision and Pattern Recognition

In Meta & Continual Learning

In Knowledge Graphs & Representation

Citations in the corpus

Incremental Prompting: Episodic Memory Prompt for Lifelong Event Detection (2022, background): “Though experience replay (Lopez-Paz and Ranzato, 2017; de Masson d’Autume et al., 2019; Guo et al., 2020; Han et al., 2020; Zhao et al., 2022) and knowledge distillation (Chuang et al.”
Continual Sequence Generation with Adaptive Compositional Modules (2022, methodology): “Besides replaying old examples, regularization on the hidden states (Wang et al., 2019; Han et al., 2020; Huang et al., 2021) or parameters (Mi et al., 2020) could be further added to prevent severe distortion.”
Continual Few-shot Relation Learning via Embedding Space Regularization and Data Augmentation (2022, methodology): “Figure 8: Relation extraction results for ERDA (our) and EMAR (Han et al., 2020) on the FewRel benchmark under the CRL setting.”
Total Recall: a Customized Continual Learning Method for Neural Semantic Parsers (2021, methodology): “The other one evaluates the test sets of all seen tasks combined after finishing model training on task T (k), ACCwhole = accD(1:k)test (Wang et al., 2019; Han et al., 2020).”
Lifelong Intent Detection via Multi-Strategy Rebalancing (2021, methodology): “Due to the simplicity and effectiveness of replay-based methods, they become an excellent solution for lifelong learning in natural language processing scenarios [2, 9].”
Continual Learning for Text Classification with Information Disentanglement Based Regularization (2021, methodology): “…on text hidden space or model parameters, it generally views them as equally important and regularize them to the same extent (Wang et al., 2019; Han et al., 2020), making it hard for models to differentiate informative representation that needs to be retained from ones that need a large…”
Curriculum-Meta Learning for Order-Robust Continual Relation Extraction (2021, methodology): “(See Appendix C for the result for EMAR, MLLRE, and CML.) As shown in Table ??, comparing the results in the rows of P0 and P9 of each task, we can observe that most tasks see a significant drop in accuracy from P9 (when the task is the last one seen by the model) to P0 (when the task is the first o”
Continual Learning in Task-Oriented Dialogue Systems (2020, methodology): “Finally, continual learning has been used for sentence encoder (Liu et al., 2019), composition language learning (Li et al., 2019c) and relation learning (Han et al., 2020).”
A Two-phase Prototypical Network Model for Incremental Few-shot Relation Classification (2020, background): “To incrementally recognize the novel relations, two kinds of solutions, i.e. re-training (Gidaris and Komodakis, 2018) and lifelong learning (Wang et al., 2019; Han et al., 2020), are proposed.”
Prompt Augmented Generative Replay via Supervised Contrastive Learning for Lifelong Intent Detection (2022, methodology): “…) model where, similar to existing approaches (de Masson d'Autume et al., 2019; Wang et al., 2020; Qian et al., 2021; Xia et al., 2021; Wang et al., 2019; Obamuyide and Vlachos, 2019; Han et al., 2020) we do not store real samples in replay memory but instead store concept words for each intent.”
Pretrained Language Model in Continual Learning: A Comparative Study (2022, methodology): “Moreover, pretrained language models (PLMs) have been widely applied in CL methods in NLP, and their addition further complicates the clear understanding of model performance (Han et al., 2020; Wang et al., 2019; Cao et al., 2020).”
Orthogonal Subspace Learning for Language Model Continual Learning (2023, background): “However, the storage and replay of data from previous tasks raise privacy concerns, particularly when dealing with sensitive information.”
Refining Sample Embeddings with Relation Prototypes to Enhance Continual Relation Extraction (2021, methodology): “In previous CRE experiments (Wang et al., 2019; Han et al., 2020), relations are ﬁrst divided into 10 clusters to simulate 10 tasks.”
Continual Few-shot Intent Detection (2022, background): “The main problem for continual text classification is catastrophic forgetting and replay-based methods (Han et al., 2020; Cui et al., 2021) have been proven promising to alleviate the problem, which retain a few examples in previous tasks and continually replay them with new tasks.”
Prompt Conditioned VAE: Enhancing Generative Replay for Lifelong Learning in Task-Oriented Dialogue (2022, background): “One line of studies (Rebufﬁ et al., 2017; Chaudhry et al., 2019; Lopez-Paz and Ranzato, 2017; Mi et al., 2020; Han et al., 2020; Liu et al., 2021b) keeps a small number of real samples from old tasks for replaying.”
Meta-Learning with Sparse Experience Replay for Lifelong Language Learning (2020, methodology): “Han et al. (2020) 2020) present a model based on GPT-2 (Radford et al., 2019), called LAMOL, that simultaneously learns to solve new tasks and to generate pseudo-samples from previous tasks for replay.”
Incremental Event Detection via Knowledge Consolidation Networks (2020, methodology): “In replay-based methods, these methods preserve the previous knowledge via storing a small number of old data (Castro et al., 2018; Wang et al., 2019a; Han et al., 2020).”
Consistent Representation Learning for Continual Relation Extraction (2022, methodology): “In recent years, the memory-based continual relation extraction model has made signiﬁcant progress in alleviating the problem of catastrophic forgetting (Han et al., 2020; Wu et al., 2021; Cui et al., 2021).”