Neural Network based models have been state-of-the-art models for various Natural Language Processing tasks, however, the input and output dimension problem in the networks has still not been fully resolved, especially in text generation tasks (e.g. Machine Translation, Text Summarization), in which input and output both have huge sizes of vocabularies. Therefore, input-output embedding weight sharing has been introduced and adopted widely, which remains to be improved. Based on linear algebra and statistical theories, this paper locates the shortcoming of existed input-output embedding weight sharing method, then raises methods for improving input-output weight shared embedding, among which methods of normalization of embedding weight matrices show best performance. These methods are nearly computational cost-free, can get combined with other embedding techniques, and show good effectiveness when applied on state-of-the-art Neural Network models. For Transformer-big models, the normalization techniques can get at best 0.6 BLEU improvement compared to the original version of model on WMT'16 En-De dataset, and similar BLEU improvements on IWSLT 14' datasets. For DynamicConv models, 0.5 BLEU improvement can be attained on WMT'16 En-De dataset, and 0.41 BLEU improvement on IWSLT 14' De-En translation task is achieved.
Story generation, namely generating a reasonable story from a leading context, is an important but challenging task. In spite of the success in modeling fluency and local coherence, existing neural language generation models (e.g., GPT-2) still suffer from repetition, logic conflicts, and lack of long-range coherence in generated stories. We conjecture that this is because of the difficulty of associating relevant commonsense knowledge, understanding the causal relationships, and planning entities and events with proper temporal order. In this paper, we devise a knowledge-enhanced pretraining model for commonsense story generation. We propose to utilize commonsense knowledge from external knowledge bases to generate reasonable stories. To further capture the causal and temporal dependencies between the sentences in a reasonable story, we employ multi-task learning which combines a discriminative objective to distinguish true and fake stories during fine-tuning. Automatic and manual evaluation shows that our model can generate more reasonable stories than state-of-the-art baselines, particularly in terms of logic and global coherence.
This article presents a stochastic corpus-based model for generating natural language text. Our model first encodes dependency relations from training data through a feature set, then concatenates these features to produce a new dependency tree for a given meaning representation, and finally generates a natural language utterance from the produced dependency tree. We test our model on nine domains from tabular, dialogue act and RDF format. Our model outperforms the corpus-based state-of-the-art methods trained on tabular datasets and also achieves comparable results with neural network-based approaches trained on dialogue act, E2E and WebNLG datasets for BLEU and ERR evaluation metrics. Also, by reporting Human Evaluation results, we show that our model produces high-quality utterances in aspects of informativeness and naturalness as well as quality.
Data-to-text generation models face challenges in ensuring data fidelity by referring to the correct input source. To inspire studies in this area, Wiseman et al. (2017) introduced the RotoWire corpus on generating NBA game summaries from the box- and line-score tables. However, limited attempts have been made in this direction and the challenges remain. We observe a prominent bottleneck in the corpus where only about 60% of the summary contents can be grounded to the boxscore records. Such information deficiency tends to misguide a conditioned language model to produce unconditioned random facts and thus leads to factual hallucinations. In this work, we restore the information balance and revamp this task to focus on fact-grounded data-to-text generation. We introduce a purified and larger-scale dataset, RotoWire-FG (Fact-Grounding), with 50% more data from the year 2017-19 and enriched input tables, hoping to attract more research focuses in this direction. Moreover, we achieve improved data fidelity over the state-of-the-art models by integrating a new form of table reconstruction as an auxiliary task to boost the generation quality.
PatentTransformer is our codename for patent text generation based on Transformer-based models. Our goal is "Augmented Inventing." In this second version, we leverage more of the structural metadata in patents. The structural metadata includes patent title, abstract, and dependent claim, in addition to independent claim previously. Metadata controls what kind of patent text for the model to generate. Also, we leverage the relation between metadata to build a text-to-text generation flow, for example, from a few words to a title, the title to an abstract, the abstract to an independent claim, and the independent claim to multiple dependent claims. The text flow can go backward because the relation is trained bidirectionally. We release our GPT-2 models trained from scratch and our code for inference so that readers can verify and generate patent text on their own. As for generation quality, we measure it by both ROUGE and Google Universal Sentence Encoder.
Paraphrase generation is a longstanding important problem in natural language processing. In addition, recent progress in deep generative models has shown promising results on discrete latent variables for text generation. Inspired by variational autoencoders with discrete latent structures, in this work, we propose a latent bag of words (BOW) model for paraphrase generation. We ground the semantics of a discrete latent variable by the BOW from the target sentences. We use this latent variable to build a fully differentiable content planning and surface realization model. Specifically, we use source words to predict their neighbors and model the target BOW with a mixture of softmax. We use Gumbel top-k reparameterization to perform differentiable subset sampling from the predicted BOW distribution. We retrieve the sampled word embeddings and use them to augment the decoder and guide its generation search space. Our latent BOW model not only enhances the decoder, but also exhibits clear interpretability. We show the model interpretability with regard to \emph{(i)} unsupervised learning of word neighbors \emph{(ii)} the step-by-step generation procedure. Extensive experiments demonstrate the transparent and effective generation process of this model.\footnote{Our code can be found at \url{https://github.com/FranxYao/dgm_latent_bow}}
Adversarial attacks against natural language processing systems, which perform seemingly innocuous modifications to inputs, can induce arbitrary mistakes to the target models. Though raised great concerns, such adversarial attacks can be leveraged to estimate the robustness of NLP models. Compared with the adversarial example generation in continuous data domain (e.g., image), generating adversarial text that preserves the original meaning is challenging since the text space is discrete and non-differentiable. To handle these challenges, we propose a target-controllable adversarial attack framework T3, which is applicable to a range of NLP tasks. In particular, we propose a tree-based autoencoder to embed the discrete text data into a continuous representation space, upon which we optimize the adversarial perturbation. A novel tree-based decoder is then applied to regularize the syntactic correctness of the generated text and manipulate it on either sentence (T3(Sent)) or word (T3(Word)) level. We consider two most representative NLP tasks: sentiment analysis and question answering (QA). Extensive experimental results and human studies show that T3 generated adversarial texts can successfully manipulate the NLP models to output the targeted incorrect answer without misleading the human. Moreover, we show that the generated adversarial texts have high transferability which enables the black-box attacks in practice. Our work sheds light on an effective and general way to examine the robustness of NLP models. Our code is publicly available at https://github.com/AI-secure/T3/.
To simultaneously capture syntax and global semantics from a text corpus, we propose a new larger-context recurrent neural network (RNN) based language model, which extracts recurrent hierarchical semantic structure via a dynamic deep topic model to guide natural language generation. Moving beyond a conventional RNN-based language model that ignores long-range word dependencies and sentence order, the proposed model captures not only intra-sentence word dependencies, but also temporal transitions between sentences and inter-sentence topic dependencies. For inference, we develop a hybrid of stochastic-gradient Markov chain Monte Carlo and recurrent autoencoding variational Bayes. Experimental results on a variety of real-world text corpora demonstrate that the proposed model not only outperforms larger-context RNN-based language models, but also learns interpretable recurrent multilayer topics and generates diverse sentences and paragraphs that are syntactically correct and semantically coherent.
Transcribing structured data into natural language descriptions has emerged as a challenging task, referred to as "data-to-text". These structures generally regroup multiple elements, as well as their attributes. Most attempts rely on translation encoder-decoder methods which linearize elements into a sequence. This however loses most of the structure contained in the data. In this work, we propose to overpass this limitation with a hierarchical model that encodes the data-structure at the element-level and the structure level. Evaluations on RotoWire show the effectiveness of our model w.r.t. qualitative and quantitative metrics.
Response generation for task-oriented dialogues involves two basic components: dialogue planning and surface realization. These two components, however, have a discrepancy in their objectives, i.e., task completion and language quality. To deal with such discrepancy, conditioned response generation has been introduced where the generation process is factorized into action decision and language generation via explicit action representations. To obtain action representations, recent studies learn latent actions in an unsupervised manner based on the utterance lexical similarity. Such an action learning approach is prone to diversities of language surfaces, which may impinge task completion and language quality. To address this issue, we propose multi-stage adaptive latent action learning (MALA) that learns semantic latent actions by distinguishing the effects of utterances on dialogue progress. We model the utterance effect using the transition of dialogue states caused by the utterance and develop a semantic similarity measurement that estimates whether utterances have similar effects. For learning semantic actions on domains without dialogue states, MsALA extends the semantic similarity measurement across domains progressively, i.e., from aligning shared actions to learning domain-specific actions. Experiments using multi-domain datasets, SMD and MultiWOZ, show that our proposed model achieves consistent improvements over the baselines models in terms of both task completion and language quality.
Transformer-based architectures represent the state of the art in sequence modeling tasks like machine translation and language understanding. Their applicability to multi-modal contexts like image captioning, however, is still largely under-explored. With the aim of filling this gap, we present M$^2$ - a Meshed Transformer with Memory for Image Captioning. The architecture improves both the image encoding and the language generation steps: it learns a multi-level representation of the relationships between image regions integrating learned a priori knowledge, and uses a mesh-like connectivity at decoding stage to exploit low- and high-level features. Experimentally, we investigate the performance of the M$^2$ Transformer and different fully-attentive models in comparison with recurrent ones. When tested on COCO, our proposal achieves a new state of the art in single-model and ensemble configurations on the "Karpathy" test split and on the online test server. We also assess its performances when describing objects unseen in the training set. Trained models and code for reproducing the experiments are publicly available at: https://github.com/aimagelab/meshed-memory-transformer.
This work-in-progress paper proposes a framework to generate and measure personalized patent claims. The objective is to help inventors conceive better inventions by learning from relevant inventors. Patent claim generation is a way of "augmented inventing." for inventors. Such patent claim generation leverages the recent transfer learning in the Deep Learning field, particularly the state-of-the-art Transformer-based models. In terms of system implementa-tion, it is planned to build an "auto-complete" function for patent claim drafting. The auto-complete function is analyzed from four different perspectives: extent of generation, generative direction, proximity of generation, and constraint in generation. Technically, the framework is composed of two Transformer models. One is for text generation and the other is for quality measurement. Specifically, the patent claim generation is based on GPT-2 model and the measurement of personalization is based on BERT model. The training data is inventor-centric and comes from the Inventors Endpoint API provided by the USPTO.
In this work, we tackle the problem of structured text generation, specifically academic paper generation in $\LaTeX{}$, inspired by the surprisingly good results of basic character-level language models. Our motivation is using more recent and advanced methods of language modeling on a more complex dataset of $\LaTeX{}$ source files to generate realistic academic papers. Our first contribution is preparing a dataset with $\LaTeX{}$ source files on recent open-source computer vision papers. Our second contribution is experimenting with recent methods of language modeling and text generation such as Transformer and Transformer-XL to generate consistent $\LaTeX{}$ code. We report cross-entropy and bits-per-character (BPC) results of the trained models, and we also discuss interesting points on some examples of the generated $\LaTeX{}$ code.
Text generation from AMR involves emitting sentences that reflect the meaning of their AMR annotations. Neural sequence-to-sequence models have successfully been used to decode strings from flattened graphs (e.g., using depth-first or random traversal). Such models often rely on attention-based decoders to map AMR node to English token sequences. Instead of linearizing AMR, we directly encode its graph structure and delegate traversal to the decoder. To enforce a sentence-aligned graph traversal and provide local graph context, we predict transition-based parser actions in addition to English words. We present two model variants: one generates parser actions prior to words, while the other interleaves actions with words.
Large transformer-based language models (LMs) trained on huge text corpora have shown unparalleled generation capabilities. However, controlling attributes of the generated language (e.g. switching topic or sentiment) is difficult without modifying the model architecture or fine-tuning on attribute-specific data and entailing the significant cost of retraining. We propose a simple alternative: the Plug and Play Language Model (PPLM) for controllable language generation, which combines a pretrained LM with one or more simple attribute classifiers that guide text generation without any further training of the LM. In the canonical scenario we present, the attribute models are simple classifiers consisting of a user-specified bag of words or a single learned layer with 100,000 times fewer parameters than the LM. Sampling entails a forward and backward pass in which gradients from the attribute model push the LM's hidden activations and thus guide the generation. Model samples demonstrate control over a range of topics and sentiment styles, and extensive automated and human annotated evaluations show attribute alignment and fluency. PPLMs are flexible in that any combination of differentiable attribute models may be used to steer text generation, which will allow for diverse and creative applications beyond the examples given in this paper.
Existing approaches for automatically generating mathematical word problems are deprived of customizability and creativity due to the inherent nature of template-based mechanisms they employ. We present a solution to this problem with the use of deep neural language generation mechanisms. Our approach uses a Character Level Long Short Term Memory Network (LSTM) to generate word problems, and uses POS (Part of Speech) tags to resolve the constraints found in the generated problems. Our approach is capable of generating Mathematics Word Problems in both English and Sinhala languages with an accuracy over 90%.
In this paper, we transform tag recommendation into a word-based text generation problem and introduce a sequence-to-sequence model. The model inherits the advantages of LSTM-based encoder for sequential modeling and attention-based decoder with local positional encodings for learning relations globally. Experimental results on Zhihu datasets illustrate the proposed model outperforms other state-of-the-art text classification based methods.
Emotional language generation is one of the keys to human-like artificial intelligence. Humans use different type of emotions depending on the situation of the conversation. Emotions also play an important role in mediating the engagement level with conversational partners. However, current conversational agents do not effectively account for emotional content in the language generation process. To address this problem, we develop a language modeling approach that generates affective content when the dialogue is situated in a given context. We use the recently released Empathetic-Dialogues corpus to build our models. Through detailed experiments, we find that our approach outperforms the state-of-the-art method on the perplexity metric by about 5 points and achieves a higher BLEU metric score.
We propose an approach towards natural language generation using a bidirectional encoder-decoder which incorporates external rewards through reinforcement learning (RL). We use attention mechanism and maximum mutual information as an initial objective function using RL. Using a two-part training scheme, we train an external reward analyzer to predict the external rewards and then use the predicted rewards to maximize the expected rewards (both internal and external). We evaluate the system on two standard dialogue corpora - Cornell Movie Dialog Corpus and Yelp Restaurant Review Corpus. We report standard evaluation metrics including BLEU, ROUGE-L, and perplexity as well as human evaluation to validate our approach.
Although n-gram language models (LMs) have been outperformed by the state-of-the-art neural LMs, they are still widely used in speech recognition due to its high efficiency in inference. In this paper, we demonstrate that n-gram LM can be improved by neural LMs through a text generation based data augmentation method. In contrast to previous approaches, we employ a large-scale general domain pre-training followed by in-domain fine-tuning strategy to construct deep Transformer based neural LMs. Large amount of in-domain text data is generated with the well trained deep Transformer to construct new n-gram LMs, which are then interpolated with baseline n-gram systems. Empirical studies on different speech recognition tasks show that the proposed approach can effectively improve recognition accuracy. In particular, our proposed approach brings significant relative word error rate reduction up to 6.0% for domains with limited in-domain data.