Language understanding in speech-based systems have attracted much attention in recent years with the growing demand for voice interface applications. However, the robustness of natural language understanding (NLU) systems to errors introduced by aut
omatic speech recognition (ASR) is under-examined. %To facilitate the research on ASR-robust general language understanding, In this paper, we propose ASR-GLUE benchmark, a new collection of 6 different NLU tasks for evaluating the performance of models under ASR error across 3 different levels of background noise and 6 speakers with various voice characteristics. Based on the proposed benchmark, we systematically investigate the effect of ASR error on NLU tasks in terms of noise intensity, error type and speaker variants. We further purpose two ways, correction-based method and data augmentation-based method to improve robustness of the NLU systems. Extensive experimental results and analysises show that the proposed methods are effective to some extent, but still far from human performance, demonstrating that NLU under ASR error is still very challenging and requires further research.
Knowledge retrieval and reasoning are two key stages in multi-hop question answering (QA) at web scale. Existing approaches suffer from low confidence when retrieving evidence facts to fill the knowledge gap and lack transparent reasoning process. In
this paper, we propose a new framework to exploit more valid facts while obtaining explainability for multi-hop QA by dynamically constructing a semantic graph and reasoning over it. We employ Abstract Meaning Representation (AMR) as semantic graph representation. Our framework contains three new ideas: (a) {tt AMR-SG}, an AMR-based Semantic Graph, constructed by candidate fact AMRs to uncover any hop relations among question, answer and multiple facts. (b) A novel path-based fact analytics approach exploiting {tt AMR-SG} to extract active facts from a large fact pool to answer questions. (c) A fact-level relation modeling leveraging graph convolution network (GCN) to guide the reasoning process. Results on two scientific multi-hop QA datasets show that we can surpass recent approaches including those using additional knowledge graphs while maintaining high explainability on OpenBookQA and achieve a new state-of-the-art result on ARC-Challenge in a computationally practicable setting.
Prior work has proved that Translation memory (TM) can boost the performance of Neural Machine Translation (NMT). In contrast to existing work that uses bilingual corpus as TM and employs source-side similarity search for memory retrieval, we propose
a new framework that uses monolingual memory and performs learnable memory retrieval in a cross-lingual manner. Our framework has unique advantages. First, the cross-lingual memory retriever allows abundant monolingual data to be TM. Second, the memory retriever and NMT model can be jointly optimized for the ultimate translation goal. Experiments show that the proposed method obtains substantial improvements. Remarkably, it even outperforms strong TM-augmented NMT baselines using bilingual TM. Owning to the ability to leverage monolingual data, our model also demonstrates effectiveness in low-resource and domain adaptation scenarios.
An important aspect of developing dialogue systems is how to evaluate and compare the performance of different systems. Existing automatic evaluation metrics are based on turn-level quality evaluation and use average scores for system-level compariso
n. In this paper, we propose to measure the performance of a dialogue system by computing the distribution-wise distance between its generated conversations and real-world conversations. Specifically, two distribution-wise metrics, FBD and PRD, are developed and evaluated. Experiments on several dialogue corpora show that our proposed metrics correlate better with human judgments than existing metrics.
Non-autoregressive generation (NAG) has recently attracted great attention due to its fast inference speed. However, the generation quality of existing NAG models still lags behind their autoregressive counterparts. In this work, we show that BERT ca
n be employed as the backbone of a NAG model to greatly improve performance. Additionally, we devise mechanisms to alleviate the two common problems of vanilla NAG models: the inflexibility of prefixed output length and the conditional independence of individual token predictions. Lastly, to further increase the speed advantage of the proposed model, we propose a new decoding strategy, ratio-first, for applications where the output lengths can be approximately estimated beforehand. For a comprehensive evaluation, we test the proposed model on three text generation tasks, including text summarization, sentence compression and machine translation. Experimental results show that our model significantly outperforms existing non-autoregressive baselines and achieves competitive performance with many strong autoregressive models. In addition, we also conduct extensive analysis experiments to reveal the effect of each proposed component.
We study the learning of a matching model for dialogue response selection. Motivated by the recent finding that models trained with random negative samples are not ideal in real-world scenarios, we propose a hierarchical curriculum learning framework
that trains the matching model in an easy-to-difficult scheme. Our learning framework consists of two complementary curricula: (1) corpus-level curriculum (CC); and (2) instance-level curriculum (IC). In CC, the model gradually increases its ability in finding the matching clues between the dialogue context and a response candidate. As for IC, it progressively strengthens the models ability in identifying the mismatching information between the dialogue context and a response candidate. Empirical studies on three benchmark datasets with three state-of-the-art matching models demonstrate that the proposed learning framework significantly improves the model performance across various evaluation metrics.
We propose the task of narrative incoherence detection as a new arena for inter-sentential semantic understanding: Given a multi-sentence narrative, decide whether there exist any semantic discrepancies in the narrative flow. Specifically, we focus o
n the missing sentence and discordant sentence detection. Despite its simple setup, this task is challenging as the model needs to understand and analyze a multi-sentence narrative, and predict incoherence at the sentence level. As an initial step towards this task, we implement several baselines either directly analyzing the raw text (textit{token-level}) or analyzing learned sentence representations (textit{sentence-level}). We observe that while token-level modeling has better performance when the input contains fewer sentences, sentence-level modeling performs better on longer narratives and possesses an advantage in efficiency and flexibility. Pre-training on large-scale data and auxiliary sentence prediction training objective further boost the detection performance of the sentence-level model.
Manipulating visual attributes of images through human-written text is a very challenging task. On the one hand, models have to learn the manipulation without the ground truth of the desired output. On the other hand, models have to deal with the inh
erent ambiguity of natural language. Previous research usually requires either the user to describe all the characteristics of the desired image or to use richly-annotated image captioning datasets. In this work, we propose a novel unsupervised approach, based on image-to-image translation, that alters the attributes of a given image through a command-like sentence such as change the hair color to black. Contrarily to state-of-the-art approaches, our model does not require a human-annotated dataset nor a textual description of all the attributes of the desired image, but only those that have to be modified. Our proposed model disentangles the image content from the visual attributes, and it learns to modify the latter using the textual description, before generating a new image from the content and the modified attribute representation. Because text might be inherently ambiguous (blond hair may refer to different shadows of blond, e.g. golden, icy, sandy), our method generates multiple stochast
For dialogue response generation, traditional generative models generate responses solely from input queries. Such models rely on insufficient information for generating a specific response since a certain query could be answered in multiple ways. Co
nsequentially, those models tend to output generic and dull responses, impeding the generation of informative utterances. Recently, researchers have attempted to fill the information gap by exploiting information retrieval techniques. When generating a response for a current query, similar dialogues retrieved from the entire training data are considered as an additional knowledge source. While this may harvest massive information, the generative models could be overwhelmed, leading to undesirable performance. In this paper, we propose a new framework which exploits retrieval results via a skeleton-then-response paradigm. At first, a skeleton is generated by revising the retrieved responses. Then, a novel generative model uses both the generated skeleton and the original query for response generation. Experimental results show that our approaches significantly improve the diversity and informativeness of the generated responses.
Many image processing tasks can be formulated as translating images between two image domains, such as colorization, super resolution and conditional image synthesis. In most of these tasks, an input image may correspond to multiple outputs. However,
current existing approaches only show very minor diversity of the outputs. In this paper, we present a novel approach to synthesize diverse realistic images corresponding to a semantic layout. We introduce a diversity loss objective, which maximizes the distance between synthesized image pairs and links the input noise to the semantic segments in the synthesized images. Thus, our approach can not only produce diverse images, but also allow users to manipulate the output images by adjusting the noise manually. Experimental results show that images synthesized by our approach are significantly more diverse than that of the current existing works and equipping our diversity loss does not degrade the reality of the base networks.
