The Softmax function is used in the final layer of nearly all existing sequence-to-sequence models for language generation. However, it is usually the slowest layer to compute which limits the vocabulary size to a subset of most frequent types; and it has a large memory footprint. We propose a general technique for replacing the softmax layer with a continuous embedding layer. Our primary innovations are a novel probabilistic loss, and a training and inference procedure in which we generate a probability distribution over pre-trained word embeddings, instead of a multinomial distribution over the vocabulary obtained via softmax. We evaluate this new class of sequence-to-sequence models with continuous outputs on the task of neural machine translation. We show that our models obtain upto 2.5x speed-up in training time while performing on par with the state-of-the-art models in terms of translation quality. These models are capable of handling very large vocabularies without compromising on translation quality. They also produce more meaningful errors than in the softmax-based models, as these errors typically lie in a subspace of the vector space of the reference translations.
We explore the performance of latent variable models for conditional text generation in the context of neural machine translation (NMT). Similar to Zhang et al., we augment the encoder-decoder NMT paradigm by introducing a continuous latent variable to model features of the translation process. We extend this model with a co-attention mechanism motivated by Parikh et al. in the inference network. Compared to the vision domain, latent variable models for text face additional challenges due to the discrete nature of language, namely posterior collapse. We experiment with different approaches to mitigate this issue. We show that our conditional variational model improves upon both discriminative attention-based translation and the variational baseline presented in Zhang et al. Finally, we present some exploration of the learned latent space to illustrate what the latent variable is capable of capturing. This is the first reported conditional variational model for text that meaningfully utilizes the latent variable without weakening the translation model.
We consider the problem of understanding real world tasks depicted in visual images. While most existing image captioning methods excel in producing natural language descriptions of visual scenes involving human tasks, there is often the need for an understanding of the exact task being undertaken rather than a literal description of the scene. We leverage insights from real world task understanding systems, and propose a framework composed of convolutional neural networks, and an external hierarchical task ontology to produce task descriptions from input images. Detailed experiments highlight the efficacy of the extracted descriptions, which could potentially find their way in many applications, including image alt text generation.
Generative Adversarial Networks (GANs) have proven to be a powerful framework for learning to draw samples from complex distributions. However, GANs are also notoriously difficult to train, with mode collapse and oscillations a common problem. We hypothesize that this is at least in part due to the evolution of the generator distribution and the catastrophic forgetting tendency of neural networks, which leads to the discriminator losing the ability to remember synthesized samples from previous instantiations of the generator. Recognizing this, our contributions are twofold. First, we show that GAN training makes for a more interesting and realistic benchmark for continual learning methods evaluation than some of the more canonical datasets. Second, we propose leveraging continual learning techniques to augment the discriminator, preserving its ability to recognize previous generator samples. We show that the resulting methods add only a light amount of computation, involve minimal changes to the model, and result in better overall performance on the examined image and text generation tasks.
In this paper, we compare various methods to compress a text using a neural model. We find that extracting tokens as latent variables significantly outperforms the state-of-the-art discrete latent variable models such as VQ-VAE. Furthermore, we compare various extractive compression schemes. There are two best-performing methods that perform equally. One method is to simply choose the tokens with the highest tf-idf scores. Another is to train a bidirectional language model similar to ELMo and choose the tokens with the highest loss. If we consider any subsequence of a text to be a text in a broader sense, we conclude that language is a strong compression code of itself. Our finding justifies the high quality of generation achieved with hierarchical method, as their latent variables are nothing but natural language summary. We also conclude that there is a hierarchy in language such that an entire text can be predicted much more easily based on a sequence of a small number of keywords, which can be easily found by classical methods as tf-idf. We speculate that this extraction process may be useful for unsupervised hierarchical text generation.
This article proposes Adversarially-Trained Normalized Noisy-Feature Auto-Encoder (ATNNFAE) for byte-level text generation. An ATNNFAE consists of an auto-encoder where the internal code is normalized on the unit sphere and corrupted by additive noise. Simultaneously, a replica of the decoder (sharing the same parameters as the AE decoder) is used as the generator and fed with random latent vectors. An adversarial discriminator is trained to distinguish training samples reconstructed from the AE from samples produced through the random-input generator, making the entire generator-discriminator path differentiable for discrete data like text. The combined effect of noise injection in the code and shared weights between the decoder and the generator can prevent the mode collapsing phenomenon commonly observed in GANs. Since perplexity cannot be applied to non-sequential text generation, we propose a new evaluation method using the total variance distance between frequencies of hash-coded byte-level n-grams (NGTVD). NGTVD is a single benchmark that can characterize both the quality and the diversity of the generated texts. Experiments are offered in 6 large-scale datasets in Arabic, Chinese and English, with comparisons against n-gram baselines and recurrent neural networks (RNNs). Ablation study on both the noise level and the discriminator is performed. We find that RNNs have trouble competing with the n-gram baselines, and the ATNNFAE results are generally competitive.
Recent deep learning models have shown improving results to natural language generation (NLG) irrespective of providing sufficient annotated data. However, a modest training data may harm such models performance. Thus, how to build a generator that can utilize as much of knowledge from a low-resource setting data is a crucial issue in NLG. This paper presents a variational neural-based generation model to tackle the NLG problem of having limited labeled dataset, in which we integrate a variational inference into an encoder-decoder generator and introduce a novel auxiliary autoencoding with an effective training procedure. Experiments showed that the proposed methods not only outperform the previous models when having sufficient training dataset but also show strong ability to work acceptably well when the training data is scarce.
Current conversational systems can follow simple commands and answer basic questions, but they have difficulty maintaining coherent and open-ended conversations about specific topics. Competitions like the Conversational Intelligence (ConvAI) challenge are being organized to push the research development towards that goal. This article presents in detail the RLLChatbot that participated in the 2017 ConvAI challenge. The goal of this research is to better understand how current deep learning and reinforcement learning tools can be used to build a robust yet flexible open domain conversational agent. We provide a thorough description of how a dialog system can be built and trained from mostly public-domain datasets using an ensemble model. The first contribution of this work is a detailed description and analysis of different text generation models in addition to novel message ranking and selection methods. Moreover, a new open-source conversational dataset is presented. Training on this data significantly improves the Recall@k score of the ranking and selection mechanisms compared to our baseline model responsible for selecting the message returned at each interaction.
Generating high-quality text with sufficient diversity is essential for a wide range of Natural Language Generation (NLG) tasks. Maximum-Likelihood (MLE) models trained with teacher forcing have consistently been reported as weak baselines, where poor performance is attributed to exposure bias (Bengio et al., 2015; Ranzato et al., 2015); at inference time, the model is fed its own prediction instead of a ground-truth token, which can lead to accumulating errors and poor samples. This line of reasoning has led to an outbreak of adversarial based approaches for NLG, on the account that GANs do not suffer from exposure bias. In this work, we make several surprising observations which contradict common beliefs. First, we revisit the canonical evaluation framework for NLG, and point out fundamental flaws with quality-only evaluation: we show that one can outperform such metrics using a simple, well-known temperature parameter to artificially reduce the entropy of the model's conditional distributions. Second, we leverage the control over the quality / diversity trade-off given by this parameter to evaluate models over the whole quality-diversity spectrum and find MLE models constantly outperform the proposed GAN variants over the whole quality-diversity space. Our results have several implications: 1) The impact of exposure bias on sample quality is less severe than previously thought, 2) temperature tuning provides a better quality / diversity trade-off than adversarial training while being easier to train, easier to cross-validate, and less computationally expensive. Code to reproduce the experiments is available at github.com/pclucas14/GansFallingShort
In this work, we address the problem of modifying textual attributes of sentences. Given an input sentence and a set of attribute labels, we attempt to generate sentences that are compatible with the conditioning information. To ensure that the model generates content compatible sentences, we introduce a reconstruction loss which interpolates between auto-encoding and back-translation loss components. We propose an adversarial loss to enforce generated samples to be attribute compatible and realistic. Through quantitative, qualitative and human evaluations we demonstrate that our model is capable of generating fluent sentences that better reflect the conditioning information compared to prior methods. We further demonstrate that the model is capable of simultaneously controlling multiple attributes.
Concept-to-text generation typically employs a pipeline architecture, which often leads to suboptimal texts. Content selection, for example, may greedily select the most important facts, which may require, however, too many words to express, and this may be undesirable when space is limited or expensive. Selecting other facts, possibly only slightly less important, may allow the lexicalization stage to use much fewer words, or to report more facts in the same space. Decisions made during content selection and lexicalization may also lead to more or fewer sentence aggregation opportunities, affecting the length and readability of the resulting texts. Building upon on a publicly available state of the art natural language generator for Semantic Web ontologies, this article presents an Integer Linear Programming model that, unlike pipeline architectures, jointly considers choices available in content selection, lexicalization, and sentence aggregation to avoid greedy local decisions and produce more compact texts, i.e., texts that report more facts per word. Compact texts are desirable, for example, when generating advertisements to be included in Web search results, or when summarizing structured information in limited space. An extended version of the proposed model also considers a limited form of referring expression generation and avoids redundant sentences. An approximation of the two models can be used when longer texts need to be generated. Experiments with three ontologies confirm that the proposed models lead to more compact texts, compared to pipeline systems, with no deterioration or with improvements in the perceived quality of the generated texts.
To help enforce data-protection regulations such as GDPR and detect unauthorized uses of personal data, we develop a new \emph{model auditing} technique that helps users check if their data was used to train a machine learning model. We focus on auditing deep-learning models that generate natural-language text, including word prediction and dialog generation. These models are at the core of popular online services and are often trained on personal data such as users' messages, searches, chats, and comments. We design and evaluate a black-box auditing method that can detect, with very few queries to a model, if a particular user's texts were used to train it (among thousands of other users). We empirically show that our method can successfully audit well-generalized models that are not overfitted to the training data. We also analyze how text-generation models memorize word sequences and explain why this memorization makes them amenable to auditing.
Generating coherent and cohesive long-form texts is a challenging task. Previous works relied on large amounts of human-generated texts to train neural language models. However, few attempted to explicitly improve neural language models from the perspectives of coherence and cohesion. In this work, we propose a new neural language model that is equipped with two neural discriminators which provide feedback signals at the levels of sentence (cohesion) and paragraph (coherence). Our model is trained using a simple yet efficient variant of policy gradient, called negative-critical sequence training, which is proposed to eliminate the need of training a separate critic for estimating baseline. Results demonstrate the effectiveness of our approach, showing improvements over the strong baseline -- recurrent attention-based bidirectional MLE-trained neural language model.
Many concept-to-text generation systems require domain-specific linguistic resources to produce high quality texts, but manually constructing these resources can be tedious and costly. Focusing on NaturalOWL, a publicly available state of the art natural language generator for OWL ontologies, we propose methods to extract from the Web sentence plans and natural language names, two of the most important types of domain-specific linguistic resources used by the generator. Experiments show that texts generated using linguistic resources extracted by our methods in a semi-automatic manner, with minimal human involvement, are perceived as being almost as good as texts generated using manually authored linguistic resources, and much better than texts produced by using linguistic resources extracted from the relation and entity identifiers of the ontology.
Generative Adversarial Networks (GANs) are a promising approach for text generation that, unlike traditional language models (LM), does not suffer from the problem of ``exposure bias''. However, A major hurdle for understanding the potential of GANs for text generation is the lack of a clear evaluation metric. In this work, we propose to approximate the distribution of text generated by a GAN, which permits evaluating them with traditional probability-based LM metrics. We apply our approximation procedure on several GAN-based models and show that they currently perform substantially worse than state-of-the-art LMs. Our evaluation procedure promotes better understanding of the relation between GANs and LMs, and can accelerate progress in GAN-based text generation.
Most previous work on neural text generation from graph-structured data relies on standard sequence-to-sequence methods. These approaches linearise the input graph to be fed to a recurrent neural network. In this paper, we propose an alternative encoder based on graph convolutional networks that directly exploits the input structure. We report results on two graph-to-sequence datasets that empirically show the benefits of explicitly encoding the input graph structure.
Many challenges in natural language processing require generating text, including language translation, dialogue generation, and speech recognition. For all of these problems, text generation becomes more difficult as the text becomes longer. Current language models often struggle to keep track of coherence for long pieces of text. Here, we attempt to have the model construct and use an outline of the text it generates to keep it focused. We find that the usage of an outline improves perplexity. We do not find that using the outline improves human evaluation over a simpler baseline, revealing a discrepancy in perplexity and human perception. Similarly, hierarchical generation is not found to improve human evaluation scores.
Generative Adversarial Networks (GANs) have experienced a recent surge in popularity, performing competitively in a variety of tasks, especially in computer vision. However, GAN training has shown limited success in natural language processing. This is largely because sequences of text are discrete, and thus gradients cannot propagate from the discriminator to the generator. Recent solutions use reinforcement learning to propagate approximate gradients to the generator, but this is inefficient to train. We propose to utilize an autoencoder to learn a low-dimensional representation of sentences. A GAN is then trained to generate its own vectors in this space, which decode to realistic utterances. We report both random and interpolated samples from the generator. Visualization of sentence vectors indicate our model correctly learns the latent space of the autoencoder. Both human ratings and BLEU scores show that our model generates realistic text against competitive baselines.
BLEU is widely considered to be an informative metric for text-to-text generation, including Text Simplification (TS). TS includes both lexical and structural aspects. In this paper we show that BLEU is not suitable for the evaluation of sentence splitting, the major structural simplification operation. We manually compiled a sentence splitting gold standard corpus containing multiple structural paraphrases, and performed a correlation analysis with human judgments. We find low or no correlation between BLEU and the grammaticality and meaning preservation parameters where sentence splitting is involved. Moreover, BLEU often negatively correlates with simplicity, essentially penalizing simpler sentences.
The knowledge graph(KG) composed of entities with their descriptions and attributes, and relationship between entities, is finding more and more application scenarios in various natural language processing tasks. In a typical knowledge graph like Wikidata, entities usually have a large number of attributes, but it is difficult to know which ones are important. The importance of attributes can be a valuable piece of information in various applications spanning from information retrieval to natural language generation. In this paper, we propose a general method of using external user generated text data to evaluate the relative importance of an entity's attributes. To be more specific, we use the word/sub-word embedding techniques to match the external textual data back to entities' attribute name and values and rank the attributes by their matching cohesiveness. To our best knowledge, this is the first work of applying vector based semantic matching to important attribute identification, and our method outperforms the previous traditional methods. We also apply the outcome of the detected important attributes to a language generation task; compared with previous generated text, the new method generates much more customized and informative messages.