PyTorchKR Papers

최신 논문

총 9,642편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1906.08885

Low-Resource Corpus Filtering using Multilingual Sentence Embeddings

In this paper, we describe our submission to the WMT19 low-resource parallel corpus filtering shared task. Our main approach is based on the LASER toolkit (Language-Agnostic SEntence Representations), which uses an encoder-decoder architecture trained on a parallel corpus to obtain multilingual sentence representations. We then use the representations directly to score and filter the noisy parallel sentences without additionally training a scoring function. We contrast our approach to other promising methods and show that LASER yields strong results. Finally, we produce an ensemble of different scoring methods and obtain additional gains. Our submission achieved the best overall performance for both the Nepali-English and Sinhala-English 1M tasks by a margin of 1.3 and 1.4 BLEU respectively, as compared to the second best systems. Moreover, our experiments show that this technique is promising for low and even no-resource scenarios.

arXiv:1906.08462

Nested Network with Two-Stream Pyramid for Salient Object Detection in Optical Remote Sensing Images

Arising from the various object types and scales, diverse imaging orientations, and cluttered backgrounds in optical remote sensing image (RSI), it is difficult to directly extend the success of salient object detection for nature scene image to the optical RSI. In this paper, we propose an end-to-end deep network called LV-Net based on the shape of network architecture, which detects salient objects from optical RSIs in a purely data-driven fashion. The proposed LV-Net consists of two key modules, i.e., a two-stream pyramid module (L-shaped module) and an encoder-decoder module with nested connections (V-shaped module). Specifically, the L-shaped module extracts a set of complementary information hierarchically by using a two-stream pyramid structure, which is beneficial to perceiving the diverse scales and local details of salient objects. The V-shaped module gradually integrates encoder detail features with decoder semantic features through nested connections, which aims at suppressing the cluttered backgrounds and highlighting the salient objects. In addition, we construct the first publicly available optical RSI dataset for salient object detection, including 800 images with varying spatial resolutions, diverse saliency types, and pixel-wise ground truth. Experiments on this benchmark dataset demonstrate that the proposed method outperforms the state-of-the-art salient object detection methods both qualitatively and quantitatively.

arXiv:1906.08745

ANAE: Learning Node Context Representation for Attributed Network Embedding

Attributed network embedding aims to learn low-dimensional node representations from both network structure and node attributes. Existing methods can be categorized into two groups: (1) the first group learns two separated node representations from network structure and node attribute respectively and concatenates them together; (2) the other group obtains node representations by translating node attributes into network structure or vice versa. However, both groups have their drawbacks. The first group neglects the correlation between network structure and node attributes, while the second group assumes strong dependence between these two types of information. In this paper, we address attributed network embedding from a novel perspective, i.e., learning node context representation for each node via modeling its attributed local subgraph. To achieve this goal, we propose a novel attributed network auto-encoder framework, namely ANAE. For a target node, ANAE first aggregates the attribute information from its attributed local subgraph, obtaining its low-dimensional representation. Next, ANAE diffuses the representation of the target node to nodes in its local subgraph to reconstruct their attributes. Such an encoder-decoder framework allows the learned representations to better preserve the context information manifested in both network structure and node attributes, thus having high capacity to learn good node representations for attributed network. Extensive experimental results on real-world datasets demonstrate that the proposed framework outperforms the state-of-the-art approaches at the tasks of link prediction and node classification.

arXiv:1906.08511공식 코드 · Papers with Code

From Zero-Shot Learning to Cold-Start Recommendation

Zero-shot learning (ZSL) and cold-start recommendation (CSR) are two challenging problems in computer vision and recommender system, respectively. In general, they are independently investigated in different communities. This paper, however, reveals that ZSL and CSR are two extensions of the same intension. Both of them, for instance, attempt to predict unseen classes and involve two spaces, one for direct feature representation and the other for supplementary description. Yet there is no existing approach which addresses CSR from the ZSL perspective. This work, for the first time, formulates CSR as a ZSL problem, and a tailor-made ZSL method is proposed to handle CSR. Specifically, we propose a Low-rank Linear Auto-Encoder (LLAE), which challenges three cruxes, i.e., domain shift, spurious correlations and computing efficiency, in this paper. LLAE consists of two parts, a low-rank encoder maps user behavior into user attributes and a symmetric decoder reconstructs user behavior from user attributes. Extensive experiments on both ZSL and CSR tasks verify that the proposed method is a win-win formulation, i.e., not only can CSR be handled by ZSL models with a significant performance improvement compared with several conventional state-of-the-art methods, but the consideration of CSR can benefit ZSL as well.

arXiv:1906.07751

Neural Volumes: Learning Dynamic Renderable Volumes from Images

Modeling and rendering of dynamic scenes is challenging, as natural scenes often contain complex phenomena such as thin structures, evolving topology, translucency, scattering, occlusion, and biological motion. Mesh-based reconstruction and tracking often fail in these cases, and other approaches (e.g., light field video) typically rely on constrained viewing conditions, which limit interactivity. We circumvent these difficulties by presenting a learning-based approach to representing dynamic objects inspired by the integral projection model used in tomographic imaging. The approach is supervised directly from 2D images in a multi-view capture setting and does not require explicit reconstruction or tracking of the object. Our method has two primary components: an encoder-decoder network that transforms input images into a 3D volume representation, and a differentiable ray-marching operation that enables end-to-end training. By virtue of its 3D representation, our construction extrapolates better to novel viewpoints compared to screen-space rendering techniques. The encoder-decoder architecture learns a latent representation of a dynamic scene that enables us to produce novel content sequences not seen during training. To overcome memory limitations of voxel-based representations, we learn a dynamic irregular grid structure implemented with a warp field during ray-marching. This structure greatly improves the apparent resolution and reduces grid-like artifacts and jagged motion. Finally, we demonstrate how to incorporate surface-based representations into our volumetric-learning framework for applications where the highest resolution is required, using facial performance capture as a case in point.

arXiv:1906.08152

Learning Disentangled Representations of Timbre and Pitch for Musical Instrument Sounds Using Gaussian Mixture Variational Autoencoders

In this paper, we learn disentangled representations of timbre and pitch for musical instrument sounds. We adapt a framework based on variational autoencoders with Gaussian mixture latent distributions. Specifically, we use two separate encoders to learn distinct latent spaces for timbre and pitch, which form Gaussian mixture components representing instrument identity and pitch, respectively. For reconstruction, latent variables of timbre and pitch are sampled from corresponding mixture components, and are concatenated as the input to a decoder. We show the model efficacy by latent space visualization, and a quantitative analysis indicates the discriminability of these spaces, even with a limited number of instrument labels for training. The model allows for controllable synthesis of selected instrument sounds by sampling from the latent spaces. To evaluate this, we trained instrument and pitch classifiers using original labeled data. These classifiers achieve high accuracy when tested on our synthesized sounds, which verifies the model performance of controllable realistic timbre and pitch synthesis. Our model also enables timbre transfer between multiple instruments, with a single autoencoder architecture, which is evaluated by measuring the shift in posterior of instrument classification. Our in depth evaluation confirms the model ability to successfully disentangle timbre and pitch.

arXiv:1906.08090공식 코드 · Papers with Code

Disentangled Inference for GANs with Latently Invertible Autoencoder

Generative Adversarial Networks (GANs) play an increasingly important role in machine learning. However, there is one fundamental issue hindering their practical applications: the absence of capability for encoding real-world samples. The conventional way of addressing this issue is to learn an encoder for GAN via Variational Auto-Encoder (VAE). In this paper, we show that the entanglement of the latent space for the VAE/GAN framework poses the main challenge for encoder learning. To address the entanglement issue and enable inference in GAN we propose a novel algorithm named Latently Invertible Autoencoder (LIA). The framework of LIA is that an invertible network and its inverse mapping are symmetrically embedded in the latent space of VAE. The decoder of LIA is first trained as a standard GAN with the invertible network and then the partial encoder is learned from a disentangled autoencoder by detaching the invertible network from LIA, thus avoiding the entanglement problem caused by the random latent space. Experiments conducted on the FFHQ face dataset and three LSUN datasets validate the effectiveness of LIA/GAN.

arXiv:1906.07947공식 코드 · Papers with Code

Generative approach to unsupervised deep local learning

Most existing feature learning methods optimize inflexible handcrafted features and the affinity matrix is constructed by shallow linear embedding methods. Different from these conventional methods, we pretrain a generative neural network by stacking convolutional autoencoders to learn the latent data representation and then construct an affinity graph with them as a prior. Based on the pretrained model and the constructed graph, we add a self-expressive layer to complete the generative model and then fine-tune it with a new loss function, including the reconstruction loss and a deliberately defined locality-preserving loss. The locality-preserving loss designed by the constructed affinity graph serves as prior to preserve the local structure during the fine-tuning stage, which in turn improves the quality of feature representation effectively. Furthermore, the self-expressive layer between the encoder and decoder is based on the assumption that each latent feature is a linear combination of other latent features, so the weighted combination coefficients of the self-expressive layer are used to construct a new refined affinity graph for representing the data structure. We conduct experiments on four datasets to demonstrate the superiority of the representation ability of our proposed model over the state-of-the-art methods.

arXiv:1906.07549

An Attention-Guided Deep Regression Model for Landmark Detection in Cephalograms

Cephalometric tracing method is usually used in orthodontic diagnosis and treatment planning. In this paper, we propose a deep learning based framework to automatically detect anatomical landmarks in cephalometric X-ray images. We train the deep encoder-decoder for landmark detection, and combine global landmark configuration with local high-resolution feature responses. The proposed frame-work is based on 2-stage u-net, regressing the multi-channel heatmaps for land-mark detection. In this framework, we embed attention mechanism with global stage heatmaps, guiding the local stage inferring, to regress the local heatmap patches in a high resolution. Besides, the Expansive Exploration strategy improves robustness while inferring, expanding the searching scope without increasing model complexity. We have evaluated our framework in the most widely-used public dataset of landmark detection in cephalometric X-ray images. With less computation and manually tuning, our framework achieves state-of-the-art results.

arXiv:1906.07480

Deep Multicameral Decoding for Localizing Unoccluded Object Instances from a Single RGB Image

Occlusion-aware instance-sensitive segmentation is a complex task generally split into region-based segmentations, by approximating instances as their bounding box. We address the showcase scenario of dense homogeneous layouts in which this approximation does not hold. In this scenario, outlining unoccluded instances by decoding a deep encoder becomes difficult, due to the translation invariance of convolutional layers and the lack of complexity in the decoder. We therefore propose a multicameral design composed of subtask-specific lightweight decoder and encoder-decoder units, coupled in cascade to encourage subtask-specific feature reuse and enforce a learning path within the decoding process. Furthermore, the state-of-the-art datasets for occlusion-aware instance segmentation contain real images with few instances and occlusions mostly due to objects occluding the background, unlike dense object layouts. We thus also introduce a synthetic dataset of dense homogeneous object layouts, namely Mikado, which extensibly contains more instances and inter-instance occlusions per image than these public datasets. Our extensive experiments on Mikado and public datasets show that ordinal multiscale units within the decoding process prove more effective than state-of-the-art design patterns for capturing position-sensitive representations. We also show that Mikado is plausible with respect to real-world problems, in the sense that it enables the learning of performance-enhancing representations transferable to real images, while drastically reducing the need of hand-made annotations for finetuning. The proposed dataset will be made publicly available.

arXiv:1906.07414

A Unified Speaker Adaptation Method for Speech Synthesis using Transcribed and Untranscribed Speech with Backpropagation

By representing speaker characteristic as a single fixed-length vector extracted solely from speech, we can train a neural multi-speaker speech synthesis model by conditioning the model on those vectors. This model can also be adapted to unseen speakers regardless of whether the transcript of adaptation data is available or not. However, this setup restricts the speaker component to just a single bias vector, which in turn limits the performance of adaptation process. In this study, we propose a novel speech synthesis model, which can be adapted to unseen speakers by fine-tuning part of or all of the network using either transcribed or untranscribed speech. Our methodology essentially consists of two steps: first, we split the conventional acoustic model into a speaker-independent (SI) linguistic encoder and a speaker-adaptive (SA) acoustic decoder; second, we train an auxiliary acoustic encoder that can be used as a substitute for the linguistic encoder whenever linguistic features are unobtainable. The results of objective and subjective evaluations show that adaptation using either transcribed or untranscribed speech with our methodology achieved a reasonable level of performance with an extremely limited amount of data and greatly improved performance with more data. Surprisingly, adaptation with untranscribed speech surpassed the transcribed counterpart in the subjective test, which reveals the limitations of the conventional acoustic model and hints at potential directions for improvements.

arXiv:1906.07330

Boosting CNN beyond Label in Inverse Problems

Convolutional neural networks (CNN) have been extensively used for inverse problems. However, their prediction error for unseen test data is difficult to estimate a priori since the neural networks are trained using only selected data and their architecture are largely considered a blackbox. This poses a fundamental challenge to neural networks for unsupervised learning or improvement beyond the label. In this paper, we show that the recent unsupervised learning methods such as Noise2Noise, Stein's unbiased risk estimator (SURE)-based denoiser, and Noise2Void are closely related to each other in their formulation of an unbiased estimator of the prediction error, but each of them are associated with its own limitations. Based on these observations, we provide a novel boosting estimator for the prediction error. In particular, by employing combinatorial convolutional frame representation of encoder-decoder CNN and synergistically combining it with the batch normalization, we provide a close form formulation for the unbiased estimator of the prediction error that can be minimized for neural network training beyond the label. Experimental results show that the resulting algorithm, what we call Noise2Boosting, provides consistent improvement in various inverse problems under both supervised and unsupervised learning setting.

arXiv:1906.07307

Towards Transfer Learning for End-to-End Speech Synthesis from Deep Pre-Trained Language Models

Modern text-to-speech (TTS) systems are able to generate audio that sounds almost as natural as human speech. However, the bar of developing high-quality TTS systems remains high since a sizable set of studio-quality <text, audio> pairs is usually required. Compared to commercial data used to develop state-of-the-art systems, publicly available data are usually worse in terms of both quality and size. Audio generated by TTS systems trained on publicly available data tends to not only sound less natural, but also exhibits more background noise. In this work, we aim to lower TTS systems' reliance on high-quality data by providing them the textual knowledge extracted by deep pre-trained language models during training. In particular, we investigate the use of BERT to assist the training of Tacotron-2, a state of the art TTS consisting of an encoder and an attention-based decoder. BERT representations learned from large amounts of unlabeled text data are shown to contain very rich semantic and syntactic information about the input text, and have potential to be leveraged by a TTS system to compensate the lack of high-quality data. We incorporate BERT as a parallel branch to the Tacotron-2 encoder with its own attention head. For an input text, it is simultaneously passed into BERT and the Tacotron-2 encoder. The representations extracted by the two branches are concatenated and then fed to the decoder. As a preliminary study, although we have not found incorporating BERT into Tacotron-2 generates more natural or cleaner speech at a human-perceivable level, we observe improvements in other aspects such as the model is being significantly better at knowing when to stop decoding such that there is much less babbling at the end of the synthesized audio and faster convergence during training.

arXiv:1906.07251

Pose Guided Fashion Image Synthesis Using Deep Generative Model

Generating a photorealistic image with intended human pose is a promising yet challenging research topic for many applications such as smart photo editing, movie making, virtual try-on, and fashion display. In this paper, we present a novel deep generative model to transfer an image of a person from a given pose to a new pose while keeping fashion item consistent. In order to formulate the framework, we employ one generator and two discriminators for image synthesis. The generator includes an image encoder, a pose encoder and a decoder. The two encoders provide good representation of visual and geometrical context which will be utilized by the decoder in order to generate a photorealistic image. Unlike existing pose-guided image generation models, we exploit two discriminators to guide the synthesis process where one discriminator differentiates between generated image and real images (training samples), and another discriminator verifies the consistency of appearance between a target pose and a generated image. We perform end-to-end training of the network to learn the parameters through back-propagation given ground-truth images. The proposed generative model is capable of synthesizing a photorealistic image of a person given a target pose. We have demonstrated our results by conducting rigorous experiments on two data sets, both quantitatively and qualitatively.

arXiv:1906.07701공식 코드 · Papers with Code

Distilling Translations with Visual Awareness

Previous work on multimodal machine translation has shown that visual information is only needed in very specific cases, for example in the presence of ambiguous words where the textual context is not sufficient. As a consequence, models tend to learn to ignore this information. We propose a translate-and-refine approach to this problem where images are only used by a second stage decoder. This approach is trained jointly to generate a good first draft translation and to improve over this draft by (i) making better use of the target language textual context (both left and right-side contexts) and (ii) making use of visual context. This approach leads to the state of the art results. Additionally, we show that it has the ability to recover from erroneous or missing words in the source language.

arXiv:1906.07689공식 코드 · Papers with Code

Expressing Visual Relationships via Language

Describing images with text is a fundamental problem in vision-language research. Current studies in this domain mostly focus on single image captioning. However, in various real applications (e.g., image editing, difference interpretation, and retrieval), generating relational captions for two images, can also be very useful. This important problem has not been explored mostly due to lack of datasets and effective models. To push forward the research in this direction, we first introduce a new language-guided image editing dataset that contains a large number of real image pairs with corresponding editing instructions. We then propose a new relational speaker model based on an encoder-decoder architecture with static relational attention and sequential multi-head attention. We also extend the model with dynamic relational attention, which calculates visual alignment while decoding. Our models are evaluated on our newly collected and two public datasets consisting of image pairs annotated with relationship sentences. Experimental results, based on both automatic and human evaluation, demonstrate that our model outperforms all baselines and existing methods on all the datasets.

arXiv:1906.07679

Multiclass segmentation as multitask learning for drusen segmentation in retinal optical coherence tomography

Automated drusen segmentation in retinal optical coherence tomography (OCT) scans is relevant for understanding age-related macular degeneration (AMD) risk and progression. This task is usually performed by segmenting the top/bottom anatomical interfaces that define drusen, the outer boundary of the retinal pigment epithelium (OBRPE) and the Bruch's membrane (BM), respectively. In this paper we propose a novel multi-decoder architecture that tackles drusen segmentation as a multitask problem. Instead of training a multiclass model for OBRPE/BM segmentation, we use one decoder per target class and an extra one aiming for the area between the layers. We also introduce connections between each class-specific branch and the additional decoder to increase the regularization effect of this surrogate task. We validated our approach on private/public data sets with 166 early/intermediate AMD Spectralis, and 200 AMD and control Bioptigen OCT volumes, respectively. Our method consistently outperformed several baselines in both layer and drusen segmentation evaluations.

arXiv:1906.06532

Attributed Graph Clustering: A Deep Attentional Embedding Approach

Graph clustering is a fundamental task which discovers communities or groups in networks. Recent studies have mostly focused on developing deep learning approaches to learn a compact graph embedding, upon which classic clustering methods like k-means or spectral clustering algorithms are applied. These two-step frameworks are difficult to manipulate and usually lead to suboptimal performance, mainly because the graph embedding is not goal-directed, i.e., designed for the specific clustering task. In this paper, we propose a goal-directed deep learning approach, Deep Attentional Embedded Graph Clustering (DAEGC for short). Our method focuses on attributed graphs to sufficiently explore the two sides of information in graphs. By employing an attention network to capture the importance of the neighboring nodes to a target node, our DAEGC algorithm encodes the topological structure and node content in a graph to a compact representation, on which an inner product decoder is trained to reconstruct the graph structure. Furthermore, soft labels from the graph embedding itself are generated to supervise a self-training graph clustering process, which iteratively refines the clustering results. The self-training process is jointly learned and optimized with the graph embedding in a unified framework, to mutually benefit both components. Experimental results compared with state-of-the-art algorithms demonstrate the superiority of our method.

arXiv:1906.06514공식 코드 · Papers with Code

PVRED: A Position-Velocity Recurrent Encoder-Decoder for Human Motion Prediction

Human motion prediction, which aims to predict future human poses given past poses, has recently seen increased interest. Many recent approaches are based on Recurrent Neural Networks (RNN) which model human poses with exponential maps. These approaches neglect the pose velocity as well as temporal relation of different poses, and tend to converge to the mean pose or fail to generate natural-looking poses. We therefore propose a novel Position-Velocity Recurrent Encoder-Decoder (PVRED) for human motion prediction, which makes full use of pose velocities and temporal positional information. A temporal position embedding method is presented and a Position-Velocity RNN (PVRNN) is proposed. We also emphasize the benefits of quaternion parameterization of poses and design a novel trainable Quaternion Transformation (QT) layer, which is combined with a robust loss function during training. We provide quantitative results for both short-term prediction in the future 0.5 seconds and long-term prediction in the future 0.5 to 1 seconds. Experiments on several benchmarks show that our approach considerably outperforms the state-of-the-art methods. In addition, qualitative visualizations in the future 4 seconds show that our approach could predict future human-like and meaningful poses in very long time horizons. Code is publicly available on GitHub: \textcolor{red}{https://github.com/hongsong-wang/PVRNN}.

arXiv:1906.06401

"My Way of Telling a Story": Persona based Grounded Story Generation

Visual storytelling is the task of generating stories based on a sequence of images. Inspired by the recent works in neural generation focusing on controlling the form of text, this paper explores the idea of generating these stories in different personas. However, one of the main challenges of performing this task is the lack of a dataset of visual stories in different personas. Having said that, there are independent datasets for both visual storytelling and annotated sentences for various persona. In this paper we describe an approach to overcome this by getting labelled persona data from a different task and leveraging those annotations to perform persona based story generation. We inspect various ways of incorporating personality in both the encoder and the decoder representations to steer the generation in the target direction. To this end, we propose five models which are incremental extensions to the baseline model to perform the task at hand. In our experiments we use five different personas to guide the generation process. We find that the models based on our hypotheses perform better at capturing words while generating stories in the target persona.