A5100671939- Advanced Image Processing Techniques
- Advanced Data Compression Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Vision and Imaging
- Video Coding and Compression Technologies
- Image and Signal Denoising Methods
- Advanced Graph Neural Networks
- Generative Adversarial Networks and Image Synthesis
- Recommender Systems and Techniques
- Distributed systems and fault tolerance
- Privacy-Preserving Technologies in Data
- Advanced Neural Network Applications
- Image Enhancement Techniques
- Image Retrieval and Classification Techniques
- Complex Network Analysis Techniques
- Digital Media Forensic Detection
- Caching and Content Delivery
- Face and Expression Recognition
- Advanced Steganography and Watermarking Techniques
- Network Security and Intrusion Detection
- Educational Technology and Pedagogy
- Tropical and Extratropical Cyclones Research
- Meteorological Phenomena and Simulations
- Advanced Clustering Algorithms Research
- Service-Oriented Architecture and Web Services
Zhejiang Meteorological Bureau
2020-2024
Nokia (Finland)
2020-2024
Nokia (Netherlands)
2024
Beijing Jiaotong University
2018-2023
National University of Defense Technology
2022-2023
Dalian University of Technology
2023
Harbin Engineering University
2023
Tencent (China)
2020-2022
Georgia Institute of Technology
2019-2021
Hengshui University
2011-2021
With the great success of graph embedding model on both academic and industry area, robustness against adversarial attack inevitably becomes a central problem in learning domain. Regardless fruitful progress, most current works perform white-box fashion: they need to access predictions labels construct their loss. However, inaccessibility real systems makes impractical system. This paper promotes frameworks more general flexible sense – we demand various kinds with black-box driven. To this...
Over recent years, deep learning-based computer vision systems have been applied to images at an ever-increasing pace, oftentimes representing the only type of consumption for those images. Given dramatic explosion in number generated per day, a question arises: how much better would image codec targeting machine-consumption perform against state-of-the-art codecs human-consumption? In this paper, we propose machines which is neural network (NN) based and end-to-end learned. particular, set...
It has been demonstrated that adversarial graphs, i.e., graphs with imperceptible perturbations added, can cause deep graph models to fail on node/graph classification tasks. In this paper, we extend the problem of community detection which is much more difficult. We focus black-box attack and aim hide targeted individuals from models, many applications in real-world scenarios, for example, protecting personal privacy social networks understanding camouflage patterns transaction networks....
Federated recommender system (FRS), which enables many local devices to train a shared model jointly without transmitting raw data, has become prevalent recommendation paradigm with privacy-preserving advantages. However, previous work on FRS performs similarity search via inner product in continuous embedding space, causes an efficiency bottleneck when the scale of items is extremely large. We argue that such scheme federated settings ignores limited capacities resource-constrained user...
As a crucial security problem, anti-spoofing in biometrics, and particularly for the face modality, has achieved great progress recent years. Still, new threats arrive inform of better, more realistic sophisticated spoofing attacks. The objective 2nd Competition on Counter Measures to 2D Face Spoofing Attacks is challenge researchers create counter measures effectively detecting variety submitted propositions are evaluated Replay-Attack database results presented this paper.
Today, according to the Cisco Annual Internet Report (2018-2023), fastest-growing category of traffic is machine-to-machine communication. In particular, communication images and videos represents a new challenge opens up perspectives in context data compression. One possible solution approach consists adapting current human-targeted image video coding standards use case machine consumption. Another developing completely compression paradigms architectures for communications. this paper, we...
The Web Services Atomic Transactions (WS-AT) specification makes it possible for businesses to engage in standard distributed transaction processing over the Internet using technology. For such business applications, trustworthy coordination of WS-AT is crucial. In this paper, we explain how render by applying Byzantine Fault Tolerance (BFT) techniques. More specifically, show protect core services described specification, namely, Activation service, Registration Completion service and...
Even though Unmanned Surface Vehicles (USVs) are increasingly used to perform various laborious and expensive offshore tasks, they still require an extensive dedicated crew supporting ensuring the safety of their operations. The recent developments in computer vision robotics further fueled interest on developing autonomous USVs that will overcome aforementioned limitations, unleashing full potential. One most vital fundamental tasks order automate ensure USV operations is water...
With the success of graph embedding model in both academic and industry areas, robustness against adversarial attack inevitably becomes a crucial problem learning. Existing works usually perform white-box fashion: they need to access predictions/labels construct their loss. However, inaccessibility makes impractical for real learning system. This paper promotes current frameworks more general flexible sense -- we consider ability various types models remain resilient black-box driven...
We present a lightweight Byzantine fault tolerance (BFT) algorithm, which can be used to render the coordination of web services business activities (WS-BA) more trustworthy. The design BFT algorithm is result comprehensive study threats WS-BA and careful analysis state model WS-BA. uses source ordering, rather than total incoming requests achieve tolerant, state-machine replication services. have implemented incorporated it into open-source Kandula framework, implements specification with...
In recent years, competitive aerobics has been rapidly popularized and developed, the level of sports skills also greatly improved. The performance some events gradually approached reached advanced level. Therefore, it is vital to invest in quantitative analysis cross-disciplinary comprehensive research related factors. This paper adopts big data technology computer vision based on convolutional neural network, according theories biomechanics image recognition, establish a loss risk...
Since vehicle attitude cannot be readily measured, this paper designs a state observer based on the information available CAN bus. The angle estimated in way is not only robust practical applications but can also replace an IMU sensor for accurate remaining fuel range prediction under complex driving conditions. primary innovation of work development extended Kalman filter (EKF)-based estimation pitch and its deployment real-world systems. Firstly, longitudinal model considering suspension...
Automatic face recognition in unconstrained environments is a challenging task. To test current trends algorithms, we organized an evaluation on mobile environment. This paper presents the results of 8 different participants using two verification metrics. Most submitted algorithms rely one or more three types features: local binary patterns, Gabor wavelet responses including phases, and color information. The best are obtained from UNILJ-ALP, which fused several image representations...
In this paper we present an end-to-end meta-learned system for image compression. Traditional machine learning based approaches to compression train one or more neural network generalization performance. However, at inference time, the encoder latent tensor output by can be optimized each test image. This optimization regarded as a form of adaptation benevolent overfitting input content. order reduce gap between training and conditions, propose new paradigm learned compression, which is on...
Graph clustering is an important technique to understand the relationships between vertices in a big graph. In this paper, we propose novel random-walk-based graph method. The proposed method restricts reach of walking agent using inflation function and normalization function. We analyze behavior limited random walk procedure algorithm for both global local problems. Previous algorithms depend on chosen fitness find clusters around seed vertex. tackles problem entirely different manner. use...
Due to the nonlinearity of artificial neural networks, designing topologies for deep convolutional networks (CNN) is a challenging task and often only heuristic approach, such as trial error, can be applied. An evolutionary algorithm solve optimization problems where fitness landscape unknown. However, algorithms are computing resource intensive, which makes it difficult when CNNs involved. In this paper, we propose an strategy find better CNNs. Incorporating concept knowledge inheritance...
In this paper, we present the design and implementation of a lightweight fault tolerance framework for Web services. With our framework, service can be rendered tolerant by replicating it across several nodes. A consensus-based algorithm is used to ensure total ordering incoming application requests replicated service, consistent membership view among replicas. The built extending an open-source WS-ReliableMessaging specification, all reliable message exchanges in conform specification. As...
Neural image coding represents now the state-of-the-art compression approach. However, a lot of work is still to be done in video domain. In this work, we propose an end-to-end learned codec that introduces several architectural novelties as well training novelties, revolving around concepts adaptation and attention. Our organized intra-frame paired with inter-frame codec. As one novelty, train model adapt motion estimation process based on resolution input video. A second novelty new neural...
Neural Network (NN)-based coding techniques are being developed for hybrid video schemes, such as the Versatile Video Coding (VVC) standard. In-loop filters and postprocessing two types of tools that aim to improve visual quality reconstructed content. These usually trained on large or image datasets with varying content, but they rarely adaptive different content types. This problem is addressed proposed content-adaptive Convolutional (CNN) post-processing filter. The approach in ways....
Typical Byzantine fault tolerance algorithms require the application requests to be executed sequentially, which may severely limit throughput of system considering that modern CPUs are equipped with multiple processing cores.In this paper, we present design and implementation a framework for software-transactional-memory based applications aims maximize concurrent while preserving strong replica consistency.The approach is on idea committing transactions according total order triggered...
Data hiding is the procedure of encoding desired information into a certain types cover media (e.g. images) to resist potential noises for data recovery, while ensuring embedded image has few perceptual perturbations. Recently, with tremendous successes gained by deep neural networks in various fields, research on learning models attracted an increasing amount attentions. In models, maximize capacity, each pixel ought be treated differently since they have different sensitivities w.r.t....
One of the core components conventional (i.e., non-learned) video codecs consists predicting a frame from previously-decoded frame, by leveraging temporal correlations. In this paper, we propose an end-to-end learned system for compressing frames. Instead relying on pixel-space motion (as with optical flow), our learns deep embeddings frames and encodes their difference in latent space. At decoder-side, attention mechanism is designed to attend space decide how different parts previous...