arXiv subjects

Xudong Zhang

Publications and source records attributed to Xudong Zhang.

63 records · Page 4Linked to original sources

arXiv2021

Retrieval and Localization with Observation Constraints

Accurate visual re-localization is very critical to many artificial intelligence applications, such as augmented reality, virtual reality, robotics and autonomous driving. To accomplish this task, we propose an integrated visual re-localization method called RLOCS by combining image retrieval, semantic consistency and geometry verification to achieve accurate estimations. The localization pipeline is designed as a coarse-to-fine paradigm. In the retrieval part, we cascade the architecture of ResNet101-GeM-ArcFace and employ DBSCAN followed by spatial verification to obtain a better initial coarse pose. We design a module called observation constraints, which combines geometry information and semantic consistency for filtering outliers. Comprehensive experiments are conducted on open datasets, including retrieval on R-Oxford5k and R-Paris6k, semantic segmentation on Cityscapes, localization on Aachen Day-Night and InLoc. By creatively modifying separate modules in the total pipeline, our method achieves many performance improvements on the challenging localization benchmarks.

cs.CV↗

arXiv2021

Binarized Aggregated Network with Quantization: Flexible Deep Learning Deployment for CSI Feedback in Massive MIMO System

Massive multiple-input multiple-output (MIMO) is one of the key techniques to achieve better spectrum and energy efficiency in 5G system. The channel state information (CSI) needs to be fed back from the user equipment to the base station in frequency division duplexing (FDD) mode. However, the overhead of the direct feedback is unacceptable due to the large antenna array in massive MIMO system. Recently, deep learning is widely adopted to the compressed CSI feedback task and proved to be effective. In this paper, a novel network named aggregated channel reconstruction network (ACRNet) is designed to boost the feedback performance with network aggregation and parametric rectified linear unit (PReLU) activation. The practical deployment of the feedback network in the communication system is also considered. Specifically, the elastic feedback scheme is proposed to flexibly adapt the network to meet different resource limitations. Besides, the network binarization technique is combined with the feature quantization for lightweight and practical deployment. Experiments show that the proposed ACRNet outperforms loads of previous state-of-the-art networks, providing a neat feedback solution with high performance, low cost and impressive flexibility.

cs.IT↗

arXiv2021

Robust Reinforcement Learning under model misspecification

Reinforcement learning has achieved remarkable performance in a wide range of tasks these days. Nevertheless, some unsolved problems limit its applications in real-world control. One of them is model misspecification, a situation where an agent is trained and deployed in environments with different transition dynamics. We propose an novel framework that utilize history trajectory and Partial Observable Markov Decision Process Modeling to deal with this dilemma. Additionally, we put forward an efficient adversarial attack method to assist robust training. Our experiments in four gym domains validate the effectiveness of our framework.

cs.LG↗

arXiv2020

FPAENet: Pneumonia Detection Network Based on Feature Pyramid Attention Enhancement

Automatic pneumonia Detection based on deep learning has increasing clinical value. Although the existing Feature Pyramid Network (FPN) and its variants have already achieved some great successes, their detection accuracies for pneumonia lesions in medical images are still unsatisfactory. In this paper, we propose a pneumonia detection network based on feature pyramid attention enhancement, which integrates attended high-level semantic features with low-level information. We add another information extracting path equipped with feature enhancement modules, which are conducted with an attention mechanism. Experimental results show that our proposed method can achieve much better performances, as a higher value of 4.02% and 3.19%, than the baselines in detecting pneumonia lesions.

eess.IV↗

arXiv2020

EPI-based Oriented Relation Networks for Light Field Depth Estimation

Light field cameras record not only the spatial information of observed scenes but also the directions of all incoming light rays. The spatial and angular information implicitly contain geometrical characteristics such as multi-view or epipolar geometry, which can be exploited to improve the performance of depth estimation. An Epipolar Plane Image (EPI), the unique 2D spatial-angular slice of the light field, contains patterns of oriented lines. The slope of these lines is associated with the disparity. Benefiting from this property of EPIs, some representative methods estimate depth maps by analyzing the disparity of each line in EPIs. However, these methods often extract the optimal slope of the lines from EPIs while ignoring the relationship between neighboring pixels, which leads to inaccurate depth map predictions. Based on the observation that an oriented line and its neighboring pixels in an EPI share a similar linear structure, we propose an end-to-end fully convolutional network (FCN) to estimate the depth value of the intersection point on the horizontal and vertical EPIs. Specifically, we present a new feature-extraction module, called Oriented Relation Module (ORM), that constructs the relationship between the line orientations. To facilitate training, we also propose a refocusing-based data augmentation method to obtain different slopes from EPIs of the same scene point. Extensive experiments verify the efficacy of learning relations and show that our approach is competitive to other state-of-the-art methods. The code and the trained models are available at https://github.com/lkyahpu/EPI_ORM.git.

cs.CV↗

arXiv2018

Monte Carlo Simulations on robustness of functional location estimator based on several functional depth

Functional data analysis has been a growing field of study in recent decades, and one fundamental task in functional data analysis is estimating the sample location. A notion called statistical depth has been extended from multivariate data to functional data, and it can provide a center-outward order for each observation within a sample of functional curves. Making use of this intuitive nature of depth methods, a depth-based trimmed mean where curves with lower depth values are excluded can be used as a robust location estimator for the sample. In this project, we first introduced several state-of-the-art depth approaches for functional data. These depths were half region depth, functional majority depth, band depth, modified band depth and functional spatial depth. Then we described a robust location estimator based on functional depth, and studied performances of these estimators based on different functional depth approaches via simulation tests. Finally, the test results showed that estimators based on functional spatial depth and modified band depth exhibited superior performances.

stat.AP↗

arXiv2016

The Covering Factor of Warm Dust in Weak Emission-Line Active Galactic Nuclei

Weak emission-line active galactic nuclei (WLAGNs) are radio-quiet active galactic nuclei (AGNs) that have nearly featureless optical spectra. We investigate the ultraviolet to mid-infrared spectral energy distributions of 73 WLAGNs (0.4<z<3) and find that most of them are similar to normal AGNs. We also calculate the covering factor of warm dust of these 73 WLAGNs. No significant difference is indicated by a KS test between the covering factor of WLAGNs and normal AGNs in the common range of bolometric luminosity. The implication for several models of WLAGNs is discussed. The super-Eddington accretion is unlikely the dominant reason for the featureless spectrum of a WLAGN. The present results favor the evolution scenario, i.e., WLAGNs are in a special stage of AGNs.

astro-ph.GA↗

arXiv2015

A General SIMD-based Approach to Accelerating Compression Algorithms

Compression algorithms are important for data oriented tasks, especially in the era of Big Data. Modern processors equipped with powerful SIMD instruction sets, provide us an opportunity for achieving better compression performance. Previous research has shown that SIMD-based optimizations can multiply decoding speeds. Following these pioneering studies, we propose a general approach to accelerate compression algorithms. By instantiating the approach, we have developed several novel integer compression algorithms, called Group-Simple, Group-Scheme, Group-AFOR, and Group-PFD, and implemented their corresponding vectorized versions. We evaluate the proposed algorithms on two public TREC datasets, a Wikipedia dataset and a Twitter dataset. With competitive compression ratios and encoding speeds, our SIMD-based algorithms outperform state-of-the-art non-vectorized algorithms with respect to decoding speeds.

cs.IR↗

arXiv2011

Battling the Internet Water Army: Detection of Hidden Paid Posters

We initiate a systematic study to help distinguish a special group of online users, called hidden paid posters, or termed "Internet water army" in China, from the legitimate ones. On the Internet, the paid posters represent a new type of online job opportunity. They get paid for posting comments and new threads or articles on different online communities and websites for some hidden purposes, e.g., to influence the opinion of other people towards certain social events or business markets. Though an interesting strategy in business marketing, paid posters may create a significant negative effect on the online communities, since the information from paid posters is usually not trustworthy. When two competitive companies hire paid posters to post fake news or negative comments about each other, normal online users may feel overwhelmed and find it difficult to put any trust in the information they acquire from the Internet. In this paper, we thoroughly investigate the behavioral pattern of online paid posters based on real-world trace data. We design and validate a new detection mechanism, using both non-semantic analysis and semantic analysis, to identify potential online paid posters. Our test results with real-world datasets show a very promising performance.

cs.SI↗