Most convolutional network (CNN)-based inpainting methods adopt standard convolution to indistinguishably treat valid pixels and holes, making them limited in handling irregular holes and more likely to generate inpainting results with color discrepancy and blurriness. Partial convolution has been suggested to address this issue, but it adopts handcrafted feature re-normalization, and only considers forward mask-updating. In this paper, we present a learnable attention map module for learning feature renormalization and mask-updating in an end-to-end manner, which is effective in adapting to irregular holes and propagation of convolution layers. Furthermore, learnable reverse attention maps are introduced to allow the decoder of U-Net to concentrate on filling in irregular holes instead of reconstructing both holes and known regions, resulting in our learnable bidirectional attention maps. Qualitative and quantitative experiments show that our method performs favorably against state-of-the-arts in generating sharper, more coherent and visually plausible inpainting results. The source code and pre-trained models will be available.
We describe a method that predicts, from a single RGB image, a depth map that describes the scene when a masked object is removed - we call this "counterfactual depth" that models hidden scene geometry together with the observations. Our method works for the same reason that scene completion works: the spatial structure of objects is simple. But we offer a much higher resolution representation of space than current scene completion methods, as we operate at pixel-level precision and do not rely on a voxel representation. Furthermore, we do not require RGBD inputs. Our method uses a standard encoder-decoder architecture, and with a decoder modified to accept an object mask. We describe a small evaluation dataset that we have collected, which allows inference about what factors affect reconstruction most strongly. Using this dataset, we show that our depth predictions for masked objects are better than other baselines.
We propose LaserTagger - a sequence tagging approach that casts text generation as a text editing task. Target texts are reconstructed from the inputs using three main edit operations: keeping a token, deleting it, and adding a phrase before the token. To predict the edit operations, we propose a novel model, which combines a BERT encoder with an autoregressive Transformer decoder. This approach is evaluated on English text on four tasks: sentence fusion, sentence splitting, abstractive summarization, and grammar correction. LaserTagger achieves new state-of-the-art results on three of these tasks, performs comparably to a set of strong seq2seq baselines with a large number of training examples, and outperforms them when the number of examples is limited. Furthermore, we show that at inference time tagging can be more than two orders of magnitude faster than comparable seq2seq models, making it more attractive for running in a live environment.
We focus on the cross-domain context-dependent text-to-SQL generation task. Based on the observation that adjacent natural language questions are often linguistically dependent and their corresponding SQL queries tend to overlap, we utilize the interaction history by editing the previous predicted query to improve the generation quality. Our editing mechanism views SQL as sequences and reuses generation results at the token level in a simple manner. It is flexible to change individual tokens and robust to error propagation. Furthermore, to deal with complex table structures in different domains, we employ an utterance-table encoder and a table-aware decoder to incorporate the context of the user utterance and the table schema. We evaluate our approach on the SParC dataset and demonstrate the benefit of editing compared with the state-of-the-art baselines which generate SQL from scratch. Our code is available at https://github.com/ryanzhumich/sparc_atis_pytorch.
Existing approaches to dialogue state tracking rely on pre-defined ontologies consisting of a set of all possible slot types and values. Though such approaches exhibit promising performance on single-domain benchmarks, they suffer from computational complexity that increases proportionally to the number of pre-defined slots that need tracking. This issue becomes more severe when it comes to multi-domain dialogues which include larger numbers of slots. In this paper, we investigate how to approach DST using a generation framework without the pre-defined ontology list. Given each turn of user utterance and system response, we directly generate a sequence of belief states by applying a hierarchical encoder-decoder structure. In this way, the computational complexity of our model will be a constant regardless of the number of pre-defined slots. Experiments on both the multi-domain and the single domain dialogue state tracking dataset show that our model not only scales easily with the increasing number of pre-defined domains and slots but also reaches the state-of-the-art performance.
In recent years, there has been remarkable progress in supervised image segmentation. Video segmentation is less explored, despite the temporal dimension being highly informative. Semantic labels, e.g. that cannot be accurately detected in the current frame, may be inferred by incorporating information from previous frames. However, video segmentation is challenging due to the amount of data that needs to be processed and, more importantly, the cost involved in obtaining ground truth annotations for each frame. In this paper, we tackle the issue of label scarcity by using consecutive frames of a video, where only one frame is annotated. We propose a deep, end-to-end trainable model which leverages temporal information in order to make use of easy to acquire unlabeled data. Our network architecture relies on a novel interconnection of two components: a fully convolutional network to model spatial information and temporal units that are employed at intermediate levels of the convolutional network in order to propagate information through time. The main contribution of this work is the guidance of the temporal signal through the network. We show that only placing a temporal module between the encoder and decoder is suboptimal (baseline). Our extensive experiments on the CityScapes dataset indicate that the resulting model can leverage unlabeled temporal frames and significantly outperform both the frame-by-frame image segmentation and the baseline approach.
Automatic recognition and segmentation methods now become the essential requirement in identifying co-seismic landslides, which are fundamental for disaster assessment and mitigation in large-scale earthquakes. This approach used to be carried out through pixel-based or object-oriented methods. However, due to the massive amount of remote sensing data, variations in different earthquake scenarios, and the efficiency requirement for post-earthquake rescue, these methods are difficult to develop into an accurate, rapid, comprehensive, and general (cross-scene) solution for co-seismic landslide recognition. This paper develops a robust model, Dense Feature Pyramid with Encoder-decoder Network (DFPENet), to understand and fuse the multi-scale features of objects in remote sensing images. The proposed method achieves a competitive segmentation accuracy on the public ISPRS 2D Semantic. Furthermore, a comprehensive and widely-used scheme is proposed for co-seismic landslide recognition, which integrates image features extracted from the DFPENet model, geologic features, temporal resolution, landslide spatial analysis, and transfer learning, while only RGB images are used. To corroborate its feasibility and applicability, the proposed scheme is applied to two earthquake-triggered landslides in Jiuzhaigou (China) and Hokkaido (Japan), using available pre- and post-earthquake remote sensing images.
The general trend in NLP is towards increasing model capacity and performance via deeper neural networks. However, simply stacking more layers of the popular Transformer architecture for machine translation results in poor convergence and high computational overhead. Our empirical analysis suggests that convergence is poor due to gradient vanishing caused by the interaction between residual connections and layer normalization. We propose depth-scaled initialization (DS-Init), which decreases parameter variance at the initialization stage, and reduces output variance of residual connections so as to ease gradient back-propagation through normalization layers. To address computational cost, we propose a merged attention sublayer (MAtt) which combines a simplified averagebased self-attention sublayer and the encoderdecoder attention sublayer on the decoder side. Results on WMT and IWSLT translation tasks with five translation directions show that deep Transformers with DS-Init and MAtt can substantially outperform their base counterpart in terms of BLEU (+1.1 BLEU on average for 12-layer models), while matching the decoding speed of the baseline model thanks to the efficiency improvements of MAtt.
We propose a Cross-lingual Encoder-Decoder model that simultaneously translates and generates sentences with Semantic Role Labeling annotations in a resource-poor target language. Unlike annotation projection techniques, our model does not need parallel data during inference time. Our approach can be applied in monolingual, multilingual and cross-lingual settings and is able to produce dependency-based and span-based SRL annotations. We benchmark the labeling performance of our model in different monolingual and multilingual settings using well-known SRL datasets. We then train our model in a cross-lingual setting to generate new SRL labeled data. Finally, we measure the effectiveness of our method by using the generated data to augment the training basis for resource-poor languages and perform manual evaluation to show that it produces high-quality sentences and assigns accurate semantic role annotations. Our proposed architecture offers a flexible method for leveraging SRL data in multiple languages.
We introduce a new classification task for scientific statements and release a large-scale dataset for supervised learning. Our resource is derived from a machine-readable representation of the arXiv.org collection of preprint articles. We explore fifty author-annotated categories and empirically motivate a task design of grouping 10.5 million annotated paragraphs into thirteen classes. We demonstrate that the task setup aligns with known success rates from the state of the art, peaking at a 0.91 F1-score via a BiLSTM encoder-decoder model. Additionally, we introduce a lexeme serialization for mathematical formulas, and observe that context-aware models could improve when also trained on the symbolic modality. Finally, we discuss the limitations of both data and task design, and outline potential directions towards increasingly complex models of scientific discourse, beyond isolated statements.
Generating paraphrases from given sentences involves decoding words step by step from a large vocabulary. To learn a decoder, supervised learning which maximizes the likelihood of tokens always suffers from the exposure bias. Although both reinforcement learning (RL) and imitation learning (IL) have been widely used to alleviate the bias, the lack of direct comparison leads to only a partial image on their benefits. In this work, we present an empirical study on how RL and IL can help boost the performance of generating paraphrases, with the pointer-generator as a base model. Experiments on the benchmark datasets show that (1) imitation learning is constantly better than reinforcement learning; and (2) the pointer-generator models with imitation learning outperform the state-of-the-art methods with a large margin.
Spoken Language Understanding (SLU) converts user utterances into structured semantic representations. Data sparsity is one of the main obstacles of SLU due to the high cost of human annotation, especially when domain changes or a new domain comes. In this work, we propose a data augmentation method with atomic templates for SLU, which involves minimum human efforts. The atomic templates produce exemplars for fine-grained constituents of semantic representations. We propose an encoder-decoder model to generate the whole utterance from atomic exemplars. Moreover, the generator could be transferred from source domains to help a new domain which has little data. Experimental results show that our method achieves significant improvements on DSTC 2\&3 dataset which is a domain adaptation setting of SLU.
Traditionally, most data-to-text applications have been designed using a modular pipeline architecture, in which non-linguistic input data is converted into natural language through several intermediate transformations. In contrast, recent neural models for data-to-text generation have been proposed as end-to-end approaches, where the non-linguistic input is rendered in natural language with much less explicit intermediate representations in-between. This study introduces a systematic comparison between neural pipeline and end-to-end data-to-text approaches for the generation of text from RDF triples. Both architectures were implemented making use of state-of-the art deep learning methods as the encoder-decoder Gated-Recurrent Units (GRU) and Transformer. Automatic and human evaluations together with a qualitative analysis suggest that having explicit intermediate steps in the generation process results in better texts than the ones generated by end-to-end approaches. Moreover, the pipeline models generalize better to unseen inputs. Data and code are publicly available.
For 3D hand and body pose estimation task in depth image, a novel anchor-based approach termed Anchor-to-Joint regression network (A2J) with the end-to-end learning ability is proposed. Within A2J, anchor points able to capture global-local spatial context information are densely set on depth image as local regressors for the joints. They contribute to predict the positions of the joints in ensemble way to enhance generalization ability. The proposed 3D articulated pose estimation paradigm is different from the state-of-the-art encoder-decoder based FCN, 3D CNN and point-set based manners. To discover informative anchor points towards certain joint, anchor proposal procedure is also proposed for A2J. Meanwhile 2D CNN (i.e., ResNet-50) is used as backbone network to drive A2J, without using time-consuming 3D convolutional or deconvolutional layers. The experiments on 3 hand datasets and 2 body datasets verify A2J's superiority. Meanwhile, A2J is of high running speed around 100 FPS on single NVIDIA 1080Ti GPU.
In this paper, we propose to guide the video caption generation with Part-of-Speech (POS) information, based on a gated fusion of multiple representations of input videos. We construct a novel gated fusion network, with one particularly designed cross-gating (CG) block, to effectively encode and fuse different types of representations, e.g., the motion and content features of an input video. One POS sequence generator relies on this fused representation to predict the global syntactic structure, which is thereafter leveraged to guide the video captioning generation and control the syntax of the generated sentence. Specifically, a gating strategy is proposed to dynamically and adaptively incorporate the global syntactic POS information into the decoder for generating each word. Experimental results on two benchmark datasets, namely MSR-VTT and MSVD, demonstrate that the proposed model can well exploit complementary information from multiple representations, resulting in improved performances. Moreover, the generated global POS information can well capture the global syntactic structure of the sentence, and thus be exploited to control the syntactic structure of the description. Such POS information not only boosts the video captioning performance but also improves the diversity of the generated captions. Our code is at: https://github.com/vsislab/Controllable_XGating.
Bidirectional Encoder Representations from Transformers (BERT) represents the latest incarnation of pretrained language models which have recently advanced a wide range of natural language processing tasks. In this paper, we showcase how BERT can be usefully applied in text summarization and propose a general framework for both extractive and abstractive models. We introduce a novel document-level encoder based on BERT which is able to express the semantics of a document and obtain representations for its sentences. Our extractive model is built on top of this encoder by stacking several inter-sentence Transformer layers. For abstractive summarization, we propose a new fine-tuning schedule which adopts different optimizers for the encoder and the decoder as a means of alleviating the mismatch between the two (the former is pretrained while the latter is not). We also demonstrate that a two-staged fine-tuning approach can further boost the quality of the generated summaries. Experiments on three datasets show that our model achieves state-of-the-art results across the board in both extractive and abstractive settings. Our code is available at https://github.com/nlpyang/PreSumm
Semi-implicit graph variational auto-encoder (SIG-VAE) is proposed to expand the flexibility of variational graph auto-encoders (VGAE) to model graph data. SIG-VAE employs a hierarchical variational framework to enable neighboring node sharing for better generative modeling of graph dependency structure, together with a Bernoulli-Poisson link decoder. Not only does this hierarchical construction provide a more flexible generative graph model to better capture real-world graph properties, but also does SIG-VAE naturally lead to semi-implicit hierarchical variational inference that allows faithful modeling of implicit posteriors of given graph data, which may exhibit heavy tails, multiple modes, skewness, and rich dependency structures. Compared to VGAE, the derived graph latent representations by SIG-VAE are more interpretable, due to more expressive generative model and more faithful inference enabled by the flexible semi-implicit construction. Extensive experiments with a variety of graph data show that SIG-VAE significantly outperforms state-of-the-art methods on several different graph analytic tasks.
In this paper, we propose a novel end-to-end deep neural network model for omnidirectional depth estimation from a wide-baseline multi-view stereo setup. The images captured with ultra wide field-of-view (FOV) cameras on an omnidirectional rig are processed by the feature extraction module, and then the deep feature maps are warped onto the concentric spheres swept through all candidate depths using the calibrated camera parameters. The 3D encoder-decoder block takes the aligned feature volume to produce the omnidirectional depth estimate with regularization on uncertain regions utilizing the global context information. In addition, we present large-scale synthetic datasets for training and testing omnidirectional multi-view stereo algorithms. Our datasets consist of 11K ground-truth depth maps and 45K fisheye images in four orthogonal directions with various objects and environments. Experimental results show that the proposed method generates excellent results in both synthetic and real-world environments, and it outperforms the prior art and the omnidirectional versions of the state-of-the-art conventional stereo algorithms.
Despite online learning (OL) techniques have boosted the performance of semi-supervised video object segmentation (VOS) methods, the huge time costs of OL greatly restrict their practicality. Matching based and propagation based methods run at a faster speed by avoiding OL techniques. However, they are limited by sub-optimal accuracy, due to mismatching and drifting problems. In this paper, we develop a real-time yet very accurate Ranking Attention Network (RANet) for VOS. Specifically, to integrate the insights of matching based and propagation based methods, we employ an encoder-decoder framework to learn pixel-level similarity and segmentation in an end-to-end manner. To better utilize the similarity maps, we propose a novel ranking attention module, which automatically ranks and selects these maps for fine-grained VOS performance. Experiments on DAVIS-16 and DAVIS-17 datasets show that our RANet achieves the best speed-accuracy trade-off, e.g., with 33 milliseconds per frame and J&F=85.5% on DAVIS-16. With OL, our RANet reaches J&F=87.1% on DAVIS-16, exceeding state-of-the-art VOS methods. The code can be found at https://github.com/Storife/RANet.
Existing conversational systems tend to generate generic responses. Recently, Background Based Conversations (BBCs) have been introduced to address this issue. Here, the generated responses are grounded in some background information. The proposed methods for BBCs are able to generate more informative responses, they either cannot generate natural responses or have difficulty in locating the right background information. In this paper, we propose a Reference-aware Network (RefNet) to address the two issues. Unlike existing methods that generate responses token by token, RefNet incorporates a novel reference decoder that provides an alternative way to learn to directly cite a semantic unit (e.g., a span containing complete semantic information) from the background. Experimental results show that RefNet significantly outperforms state-of-the-art methods in terms of both automatic and human evaluations, indicating that RefNet can generate more appropriate and human-like responses.