Conventional natural language process (NLP) generation models are trained offline with a given dataset for a particular task, which is referred to as isolated learning. Research on sequence-to-sequence language generation aims to study continual learning model to constantly learning from sequentially encountered tasks. However, continual learning studies often suffer from catastrophic forgetting, a persistent challenge for lifelong learning. In this paper, we present a novel NLP transformer model that attempts to mitigate catastrophic forgetting in online continual learning from a new perspective, i.e., attention calibration. We model the attention in the transformer as a calibrated unit in a general formulation, where the attention calibration could give benefits to balance the stability and plasticity of continual learning algorithms through influencing both their forward inference path and backward optimization path. Our empirical experiments, paraphrase generation and dialog response generation, demonstrate that this work outperforms state-of-the-art models by a considerable margin and effectively mitigate the forgetting.

Paraphrase generation is an important language generation task attempting to interpret user intents and systematically generate new phrases of identical meanings to the given ones. However, the effectiveness of paraphrase generation is constrained by the access to the golden labeled data pairs where both the amount and the quality of the training data pairs are restricted. In this paper, we propose a new weakly supervised paraphrase generation approach that extends the success of a recent work that leverages reinforcement learning for effective model training with data selection. While data selection is privileged for the target task which has noisy data, developing a reinforced selective learning regime faces several unresolved challenges. In this paper, we carry on important discussions about the above problem and present a new model that could partially overcome the discussed issues with a model-based planning feature and a reward normalization feature. We perform extensive evaluation on four weakly supervised paraphrase generation tasks where the results show that our method could significantly improve the state-of-the-art performance on the evaluation datasets.

Seq2seq language generation models that are trained offline with multiple domains in a sequential fashion often suffer from catastrophic forgetting. Lifelong learning has been proposed to handle this problem. However, existing work such as experience replay or elastic weighted consolidation requires incremental memory space. In this work, we propose an innovative framework, RMR_DSEthat leverages a recall optimization mechanism to selectively memorize important parameters of previous tasks via regularization, and uses a domain drift estimation algorithm to compensate the drift between different do-mains in the embedding space. These designs enable the model to be trained on the current task while keep-ing the memory of previous tasks, and avoid much additional data storage. Furthermore, RMR_DSE can be combined with existing lifelong learning approaches. Our experiments on two seq2seq language generation tasks, paraphrase and dialog response generation, show thatRMR_DSE outperforms SOTA models by a considerable margin and reduces forgetting greatly.

Recently, prompt learning has received significant attention, where the downstream tasks are reformulated to the mask-filling task with the help of a textual prompt. The key point of prompt learning is finding the most appropriate prompt. This paper proposes a novel model PromptGen, which can automatically generate prompts conditional on the input sentence. PromptGen is the first work considering dynamic prompt generation for knowledge probing, based on a pre-trained generative model. To mitigate any label information leaking from the pre-trained generative model, when given a generated prompt, we replace the query input with “None”. We pursue that this perturbed context-free prompt cannot trigger the correct label. We evaluate our model on the knowledge probing LAMA benchmark, and show that PromptGen significantly outperforms other baselines.

When we execute the typical fine-tuning paradigm on continuously sequential tasks, the model will suffer from the catastrophic forgetting problem (i.e., the model tends to adjust old parameters according to the new knowledge, which leads to the loss of previously acquired concepts). People proposed replay-based methods by accessing old data from extra storage and maintaining the parameters of old concepts, which actually raise the privacy issue and larger memory requirements. In this work, we aim to achieve the sequential/continual learning of knowledge without accessing the old data. The core idea is to calibrate the parameters and logits (output) so that preserving old parameters and generalized learning on new concepts can be solved simultaneously. Our proposed framework includes two major components, Logits Calibration (LC) and Parameter Calibration (PC). The LC focuses on calibrating the learning of novel models with old models, and PC aims to preserve the parameters of old models. These two operations can maintain the old knowledge while learning new tasks without storing previous data. We conduct experiments on various scenarios of the GLUE (the General Language Understanding Evaluation) benchmark. The experimental results show that our model achieves state-of-the-art performance in all scenarios.

Recently, disentanglement based on a generative adversarial network or a variational autoencoder has significantly advanced the performance of diverse applications in CV and NLP domains. Nevertheless, those models still work on coarse levels in the disentanglement of closely related properties, such as syntax and semantics in human languages. This paper introduces a deep decomposable model based on VAE to disentangle syntax and semantics by using total correlation penalties on KL divergences. Notably, we decompose the KL divergence term of the original VAE so that the generated latent variables can be separated in a more clear-cut and interpretable way. Experiments on benchmark datasets show that our proposed model can significantly improve the disentanglement quality between syntactic and semantic representations for semantic similarity tasks and syntactic similarity tasks.

For voice assistants like Alexa, Google Assistant, and Siri, correctly interpreting users’ intentions is of utmost importance. However, users sometimes experience friction with these assistants, caused by errors from different system components or user errors such as slips of the tongue. Users tend to rephrase their queries until they get a satisfactory response. Rephrase detection is used to identify the rephrases and has long been treated as a task with pairwise input, which does not fully utilize the contextual information (e.g. users’ implicit feedback). To this end, we propose a contextual rephrase detection model ContReph to automatically identify rephrases from multi-turn dialogues. We showcase how to leverage the dialogue context and user-agent interaction signals, including the user’s implicit feedback and the time gap between different turns, which can help significantly outperform the pairwise rephrase detection models.

Paraphrase generation is a longstanding NLP task that has diverse applications on downstream NLP tasks. However, the effectiveness of existing efforts predominantly relies on large amounts of golden labeled data. Though unsupervised endeavors have been proposed to alleviate this issue, they may fail to generate meaningful paraphrases due to the lack of supervision signals. In this work, we go beyond the existing paradigms and propose a novel approach to generate high-quality paraphrases with data of weak supervision. Specifically, we tackle the weakly-supervised paraphrase generation problem by: (1) obtaining abundant weakly-labeled parallel sentences via retrieval-based pseudo paraphrase expansion; and (2) developing a meta-learning framework to progressively select valuable samples for fine-tuning a pre-trained language model BART on the sentential paraphrasing task. We demonstrate that our approach achieves significant improvements over existing unsupervised approaches, and is even comparable in performance with supervised state-of-the-arts.

Text classification is a critical research topic with broad applications in natural language processing. Recently, graph neural networks (GNNs) have received increasing attention in the research community and demonstrated their promising results on this canonical task. Despite the success, their performance could be largely jeopardized in practice since they are: (1) unable to capture high-order interaction between words; (2) inefficient to handle large datasets and new documents. To address those issues, in this paper, we propose a principled model – hypergraph attention networks (HyperGAT), which can obtain more expressive power with less computational consumption for text representation learning. Extensive experiments on various benchmark datasets demonstrate the efficacy of the proposed approach on the text classification task.

Recent neural network models have significantly advanced the task of coreference resolution. However, current neural coreference models are usually trained with heuristic loss functions that are computed over a sequence of local decisions. In this paper, we introduce an end-to-end reinforcement learning based coreference resolution model to directly optimize coreference evaluation metrics. Specifically, we modify the state-of-the-art higher-order mention ranking approach in Lee et al. (2018) to a reinforced policy gradient model by incorporating the reward associated with a sequence of coreference linking actions. Furthermore, we introduce maximum entropy regularization for adequate exploration to prevent the model from prematurely converging to a bad local optimum. Our proposed model achieves new state-of-the-art performance on the English OntoNotes v5.0 benchmark.

Leveraging domain knowledge is an effective strategy for enhancing the quality of inferred low-dimensional representations of documents by topic models. In this paper, we develop topic modeling with knowledge graph embedding (TMKGE), a Bayesian nonparametric model to employ knowledge graph (KG) embedding in the context of topic modeling, for extracting more coherent topics. Specifically, we build a hierarchical Dirichlet process (HDP) based model to flexibly borrow information from KG to improve the interpretability of topics. An efficient online variational inference method based on a stick-breaking construction of HDP is developed for TMKGE, making TMKGE suitable for large document corpora and KGs. Experiments on three public datasets illustrate the superior performance of TMKGE in terms of topic coherence and document classification accuracy, compared to state-of-the-art topic modeling methods.