检索结果-内蒙古大学图书馆

Layer-wise coordination between encoder and decoder for neural machine translation 18

学校读者我要写书评

暂无评论

Layer-wise coordination between encoder and decoder for neur...

Proceedings of the 32nd International Conference on Neural information processing systems

作者： Tianyu He Xu Tan Yingce Xia Di He Tao Qin Zhibo Chen Tie-Yan Liu CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Microsoft Research Key Laboratory of Machine Perception MOE School of EECS Peking University

Neural Machine Translation (NMT) has achieved remarkable progress with the quick evolvement of model structures. In this paper, we propose the concept of layer-wise coordination for NMT, which explicitly coordinates the learning of hidden representations of the encoder and decoder together layer by layer, gradually from low level to high level. Specifically, we design a layer-wise attention and mixed attention mechanism, and further share the parameters of each layer between the encoder and decoder to regularize and coordinate the learning. Experiments show that combined with the state-of-the-art Transformer model, layer-wise coordination achieves improvements on three IWSLT and two WMT translation tasks. More specifically, our method achieves 34.43 and 29.01 BLEU score on WMT16 English-Romanian and WMT14 English-German tasks, outperforming the Transformer baseline.

关键词：

Visual analyses of music history: A user-centric approach

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Zhang, Jingxian Liu, Dong Media Lab Massachusetts Institute of Technology CambridgeMA02139 United States CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China

Music history, referring to the records of users' listening or downloading history in online music services, is the primary source for music service providers to analyze users' preferences on music and thus to provide personalized recommendations to users. In order to engage users into the service and to improve user experience, it would be beneficial to provide visual analyses of one user's music history as well as visualized recommendations to that user. In this paper, we take a user-centric approach to the design of such visual analyses. We start by investigating user needs on such visual analyses and recommendations, then propose several different visualization schemes, and perform a pilot study to collect user feedback on the designed schemes. We further conduct user studies to verify the utility of the proposed schemes, and the results not only demonstrate the effectiveness of our proposed visualization, but also provide important insights to guide the visualization design in the future. Copyright © 2017, The Authors. All rights reserved.

关键词： Visualization

cascaded Regression Tracking: Towards Online Hard Distractor Discrimination

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Wang, Ning Zhou, Wengang Tian, Qi Li, Houqiang CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System Department of Electronic Engineering and Information Science University of Science and Technology of China Hefei China Huawei Noah’s Ark Laboratory

Visual tracking can be easily disturbed by similar surrounding objects. Such objects as hard distractors, even though being the minority among negative samples, increase the risk of target drift and model corruption, which deserve additional attention in online tracking and model update. To enhance the tracking robustness, in this paper, we propose a cascaded regression tracker with two sequential stages. In the first stage, we filter out abundant easily-identified negative candidates via an efficient convolutional regression. In the second stage, a discrete sampling based ridge regression is designed to double-check the remaining ambiguous hard samples, which serves as an alternative of fully-connected layers and benefits from the closed-form solver for efficient learning. During the model update, we utilize the hard negative mining technique and an adaptive ridge regression scheme to improve the discrimination capability of the second-stage regressor. Extensive experiments are conducted on 11 challenging tracking benchmarks including OTB-2013, OTB-2015, VOT2018, VOT2019, UAV123, Temple-Color, NfS, TrackingNet, LaSOT, UAV20L, and OxUvA. The proposed method achieves state-of-the-art performance on prevalent benchmarks, while running in a real-time speed. Copyright © 2020, The Authors. All rights reserved.

关键词： Regression analysis

Copyright protection in named data networking

学校读者我要写书评

暂无评论

International Conference on Wireless Communications and Signal processing (WCSP)

作者： Xiaobin Tan Zifei Zhou Cliff Zou Yukun Niu Xin Chen CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China China Department of Electrical Engineering and Computer Science University of Central Florida USA

As a new Internet architecture, Named Data Networking (NDN) decouples location from the data itself to achieve security, scalability, and mobility. Although router-side data caching used in NDN reduces data acquisition delay, it introduces a new copyright protection challenge: how to prevent unauthorized users to retrieve data cached in routers that are out of the control of its publisher? Current approaches that rely on a common encryption key among authorized users cannot protect copyright well since if one authorized user secretly leaks the key, we cannot tell who has leaked the key out. In this paper, we present a split-based scheme to solve this copyright protection problem for large-sized data. The data is split into a large part that could be cached in routers for all users to retrieve, and a small part that is unique for each authorized user. This scheme exploits the fact that in the bit-wise OR operation, both bit 0 and bit 1 can OR with 1 to generate the same result of bit 1. The analysis of our scheme shows that it has a good performance in terms of copyright protection, data retrieval efficiency, and overhead.

关键词： Copyright protection Encryption Watermarking Internet Computer architecture Receivers

Efficient background picture coding for videos obtained from static cameras

学校读者我要写书评

暂无评论

Efficient background picture coding for videos obtained from...

IEEE Visual Communications and Image processing (VCIP)

作者： Fangdong Chen Li Li Dong Liu Houqiang Li Zhuoyi Lv Haitao Yang CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Hefei China Huawei Technologies Co. Ltd. Shenzhen China

With the exponential growth of surveillance videos, conference videos and sports videos, videos with static cameras present an unprecedented challenge for high-efficiency video coding technology. The existing schemes developed for these videos mostly encode the background as the long-term reference (LTR) to further improve the coding efficiency. However, since the bit allocation of the long-term background reference is not intensively studied, the coding efficiency is still unsatisfactory. Based on the stability analysis of the video content, an efficient background picture coding algorithm for videos obtained from static cameras, which is embedded with the basic unit level bit allocation, is proposed in this paper. Experimental results reveal that on top of the default mode in HEVC, our method offers the performance with 10.8% BD-rate reduction on average. Compared with the state-of-the-art algorithm, it still outperforms for kinds of test sequences with negligible increases of computational complexity in both encoder and decoder.

关键词： Encoding Videos Cameras Distortion Stability analysis Video coding Delays

No-reference quality assessment for 360-degree images by analysis of multi-frequency information and local-global naturalness

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Zhou, Wei Xu, Jiahua Jiang, Qiuping Chen, Zhibo CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China School of Information Science and Engineering Ningbo University Ningbo315211 China

360-degree/omnidirectional images (OIs) have achieved remarkable attentions due to the increasing applications of virtual reality (VR). Compared to conventional 2D images, OIs can provide more immersive experience to consumers, benefitting from the higher resolution and plentiful field of views (FoVs). Moreover, observing OIs is usually in the head mounted display (HMD) without references. Therefore, an efficient blind quality assessment method, which is specifically designed for 360-degree images, is urgently desired. In this paper, motivated by the characteristics of the human visual system (HVS) and the viewing process of VR visual contents, we propose a novel and effective no-reference omnidirectional image quality assessment (NR OIQA) algorithm by Multi-Frequency information and Local-Global Naturalness (MFILGN). Specifically, inspired by the frequency-dependent property of visual cortex, we first decompose the projected equirectangular projection (ERP) maps into wavelet subbands by using discrete Haar wavelet transform (DHWT). Then, the entropy intensities of low-frequency and high-frequency subbands are exploited to measure the multi-frequency information of OIs. Besides, except for considering the global naturalness of ERP maps, owing to the browsed FoVs, we extract the natural scene statistics (NSS) features from each viewport image as the measure of local naturalness. With the proposed multi-frequency information measurement and local-global naturalness measurement, we utilize support vector regression (SVR) as the final image quality regressor to train the quality evaluation model from visual quality-related features to human ratings. To our knowledge, the proposed model is the first no-reference quality assessment method for 360-degreee images that combines multi-frequency information and image naturalness. Experimental results on two publicly available OIQA databases demonstrate that our proposed MFILGN outperforms state-of-the-art full-reference (FR) a

关键词： Helmet mounted displays

Cut-and-Fold: Automatic 3D modeling from a single image

学校读者我要写书评

暂无评论

Cut-and-Fold: Automatic 3D modeling from a single image

IEEE International Conference on Multimedia and Expo Workshops (ICMEW)

作者： Guihang Wang Xuejin Chen Si Chen CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Hefei Anhui China University of California San Diego La Jolla California USA

ISBN: (纸本)9781479947164

This paper presents a novel approach to automatically generate a plausible 3D model from a single outdoor image. Outdoor scene can be modeled as a set of planar billboards such as ground, sky and vertical trees or buildings. This Cut-and-Fold system takes the advantages of the statistical classification method for high-level geometric labeling of the image regions and the automatic image segmentation algorithm for detailed boundaries of the geometric regions to generate accurate geometric labeling and realistic rendering of novel views. Taking a single image as input, our method first roughly classifies each sub-region to the geometric categories: vertical, ground or sky. Then image segmentation algorithm is employed to refine the labeling results and generate accurate boundaries for the geometric regions. After cutting the image into several regions with their geometric labels, the system folds the regions into 3D billboards with accurate boundaries. The experimental results demonstrate that our method is capable of creating plausible 3D models automatically for a wide range of outdoor scenes.

关键词： Image segmentation Abstracts labeling Indium phosphide Three-dimensional displays Solid modeling

A Close Look at Few-shot Real Image Super-resolution from the Distortion Relation Perspective

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Li, Xin Jin, Xin Fu, Jun Yu, Xiaoyuan Tong, Bei Chen, Zhibo The CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China The Eastern Institute for Advanced Study China Huawei Cloud China

Collecting amounts of distorted/clean image pairs in the real world is non-trivial, which seriously limits the practical applications of these supervised learning-based methods on real-world image super-resolution (RealSR). Previous works usually address this problem by leveraging unsupervised learning-based technologies to alleviate the dependency on paired training samples. However, these methods typically suffer from unsatisfactory texture synthesis due to the lack of supervision of clean images. To overcome this problem, we are the first to have a close look at the under-explored direction for RealSR, i.e., few-shot real-world image super-resolution, which aims to tackle the challenging RealSR problem with few-shot distorted/clean image pairs. Under this brand-new scenario, we propose Distortion Relation guided Transfer Learning (DRTL) for the few-shot RealSR by transferring the rich restoration knowledge from auxiliary distortions (i.e., synthetic distortions) to the target RealSR under the guidance of distortion relation. Concretely, DRTL builds a knowledge graph to capture the distortion relation between auxiliary distortions and target distortion (i.e., real distortions in RealSR). Based on the distortion relation, DRTL adopts a gradient reweighting strategy to guide the knowledge transfer process between auxiliary distortions and target distortions. In this way, DRTL could quickly learn the most relevant knowledge from the synthetic distortions for the target distortion. We instantiate DRTL with two commonly-used transfer learning paradigms, including pre-training and meta-learning pipelines, to realize a distortion relation-aware Few-shot RealSR. Extensive experiments on multiple benchmarks and thorough ablation studies demonstrate the effectiveness of our DRTL. Copyright © 2021, The Authors. All rights reserved.

关键词： Textures

Towards Hybrid-Optimization Video Coding

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Huo, Shuai Liu, Dong Li, Li Ma, Siwei Wu, Feng Gao, Wen The Cas Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China The Institute of Digital Media Peking University Beijing100871 China

Video coding is a mathematical optimization problem of rate and distortion essentially. To solve this complex optimization problem in practice, two popular video coding frameworks have been developed: block-based hybrid video coding and end-to-end learned video coding. If we rethink video coding from the perspective of optimization, we find that the existing two frameworks represent two directions of optimization solutions. Block-based hybrid video coding represents the discrete optimization solution because those irrelevant coding modes are discrete in mathematics. The discrete solution provides multiple starting points (i.e. modes) in global optimization space and then searches for the best one among them. However, the search-based optimization algorithm is not efficient enough. On the other hand, end-to-end learned video coding represents the continuous optimization solution because the optimization algorithm of deep learning, gradient descent, is based on a continuous function. The continuous solution optimizes a group of model parameters efficiently by such a numerical algorithm. However, limited by only one starting point, it is easy to fall into the local optimum. To better solve the optimization problem, we propose a hybrid of discrete and continuous optimization video coding. We regard video coding as a hybrid of the discrete and continuous optimization problem, and use both search and numerical algorithm to solve it. Our idea is to provide multiple discrete starting points in the global space and optimize the local optimum around each point by numerical algorithm efficiently. Finally, we search for the global optimum among those local optimums. Guided by the hybrid optimization idea, we design a hybrid optimization video coding framework, which is built on continuous deep networks entirely and also contains some discrete modes. We conduct a comprehensive set of experiments to verify the efficiency of our hybrid optimization. Compared to the continuous opti

关键词： Deep neural networks