A5100323277- Privacy-Preserving Technologies in Data
- Cryptography and Data Security
- Coding theory and cryptography
- Blockchain Technology Applications and Security
- Cryptographic Implementations and Security
- Energy Efficient Wireless Sensor Networks
- Indoor and Outdoor Localization Technologies
- Internet Traffic Analysis and Secure E-voting
- graph theory and CDMA systems
- Advanced Computational Techniques and Applications
- Network Security and Intrusion Detection
- IoT and Edge/Fog Computing
- Caching and Content Delivery
- Adversarial Robustness in Machine Learning
- Peer-to-Peer Network Technologies
- Cloud Computing and Resource Management
- Advanced Graph Neural Networks
- Cloud Data Security Solutions
- AI in cancer detection
- Digital and Cyber Forensics
- Advanced MIMO Systems Optimization
- Advanced Measurement and Detection Methods
- Complex Network Analysis Techniques
- Cooperative Communication and Network Coding
- Neural Networks and Applications
Chinese Academy of Sciences
2017-2025
Institute of Genetics and Developmental Biology
2025
Wuhan University
2010-2024
Beijing Jiaotong University
2019-2024
Guangzhou University
2019-2024
Zhejiang Lab
2023
China Electronic Information Industry Development
2023
Inspur (China)
2023
Huawei Technologies (China)
2022-2023
Tsinghua University
2011-2023
Automatic emotion recognition from speech, which is an important and challenging task in the field of affective computing, heavily relies on effectiveness speech features for classification.Previous approaches to have mostly focused extraction carefully hand-crafted features.How model spatio-temporal dynamics effectively still under active investigation.In this paper, we propose a method tackle problem emotional relevant feature by leveraging Attention-based Bidirectional Long Short-Term...
While Internet-of-Things (IoT) significantly facilitates the convenience of people's daily life, lack security practice raises risk privacy-sensitive user data leakage. Securing transmission among IoT devices is therefore a critical capability environments such as Intelligent Connected Vehicles, Smart Home, City and so forth. However, cryptographic communication scheme challenged by limited resource low-cost devices, even negligible extra CPU usage battery-powered sensors would result in...
With the growth of participating clients, centralized parameter server (PS) will seriously limit scale and efficiency Federated Learning (FL). A straightforward approach to up FL system is construct a Parallel (PFL) with multiple PSes. However, it unclear whether PFL can really achieve faster convergence rate or not. Even if answer yes, non-trivial design highly efficient average algorithm for system. In this paper, we propose completely parallelizable called P-FedAvg under architecture....
Using X-ray diffraction combined with Rietveld method and chemical analysis hydrofluoric acid (HF) solution, the content composition of glass phase a class F fly ash are quantified, initial Si/Al ratio is calculated as well. The dissolution Si4+ Al3+ in different solution (HF, NaOH, NaF) various concentrations reaction time NaOH was studied. dissolved compared value, following conclusions drawn: (1) only very small part Si Al knowing available Al; (2) influenced by alkali concentration. When...
We propose a simple yet effective method to reduce the redundancy of DenseNet by substantially decreasing number stacked modules replacing original bottleneck our SMG module, which is augmented local residual. Furthermore, module equipped with an efficient two-stage pipeline, aims DenseNet-like architectures that need integrate all previous outputs, i.e., squeezing incoming informative but redundant features gradually hierarchical convolutions as hourglass shape and then exciting it...
Aiming to the vision of 6G, integrated sensing and communication (ISAC), as potential techniques deal with increasing demands native perception ability, has recently attracted a lot attention. In order guide system optimization evaluate novel ISAC techniques, accurate realistic wireless channel model for is essential. This article proposes framework modeling extended based on 3GPP model, which can be used simulation performance analysis 6G techniques. We first overview principles...
An increasing number of data centers today start to incorporate renewable energy solutions cap their carbon footprint. However, the impact on large-scale center design is still not well understood. In this paper, we model and evaluate driven by intermittent energy. Using real-world source traces, show that power utilization load tuning frequency are two critical metrics for designing sustainable high-performance centers. Our characterization reveals fluctuation together with supply introduce...
With large available continuous bandwidth, millimeter wave (mmWave) bands hold promise as a carrier frequency for fifth generation (5G) wireless communications. Moreover, mmWave also play an important role in radar detections. Based on this observation, we propose communication and detection integrated network architecture railways, not only to increase the capacity of railway systems, but realize train operation environment enhance safety. To overcome aggravated path loss bands, directional...
With the rapid development and widespread use of wearable wireless sensors, data aggregation technique becomes one most important research areas. However, sensitive collected by sensor nodes may be leaked at intermediate aggregator nodes. So, privacy preservation is becoming an increasingly issue in security aggregation. In this paper, we propose a privacy-preserving model, which adopts mixed structure. Data integrity verified both cluster head base station. Some adopt slicing technology to...
Federated collaborative filtering (Fed-CF) is a variant of federated learning (FL) models, which can protect user privacy in recommender systems. In Fed-CF, the recommendation model collectively trained across multiple decentralized clients by exchanging gradients only. However, nature Fed-CF makes it vulnerable to shilling attacks, be realized inserting fake ratings target items distort results. Unfortunately, previous detection algorithms cannot work well FL framework, as all original data...
The advent of the big data era has brought unprecedented demands. integration computing resources with network in force enables possibility distributed collaborative training. However, unencrypted training is vulnerable to threats such as gradient inversion attacks and model theft. To address this issue, are usually protected by cryptographic methods. semantic meaninglessness encrypted makes it difficult prevent potential poisoning free-riding attacks. In paper, we propose a fairness...
In this paper, we present WoSense, a device-free and real-time behavior analysis system leveraging only WiFi infrastructures. WoSense aims to remotely recognize various human behaviors like surfing, gaming working around computers, which are considered be an essential part of our daily lives both at work home. The key is exploit the signal distortions on channel data caused by gestures finger hand movements, then identify possible via composite gestures. Therefore, two critical challenges...
Federated learning (FL) is an emerging paradigm through which decentralized devices can collaboratively train a common model. However, serious concern the leakage of privacy from exchanged gradient information between clients and parameter server (PS) in FL. To protect information, adopt differential (DP) to add additional noises distort original gradients before they are uploaded PS. Nevertheless, model accuracy will be significantly impaired by DP noises, making impracticable real systems....
The probability distribution of probabilistic shaping 64 quadrature amplitude modulation (PS-64QAM) is uneven. traditional M-ary support vector machine (SVM) algorithm incompatible with the data set uneven distribution. In order to solve problem, we propose a novel nonlinear equalizer (NLE) for PS-64QAM based on constellation segmentation (CS) and SVM, called CS SVM NLE. performance NLE has been demonstrated in 120 Gb/s coherent optical communication system. experimental results show that...
With the development of high-efficiency solar cells, issue industrial Si photovoltaic (PV) cells' light-induced degradation (LID) became even more serious. In this paper, a method for quick LID treatments was developed. A high-intensity monochromatic light source used to realize treatment, and 940-nm light-emitting diodes (LEDs) were selected build treatment platform. The photon flux density output from can reach 1.32 × 1023 m−2 s−1. From simulation calculation, photons 1.27× s−1 be utilized...
Recently, Cross-Modal Hamming space Retrieval (CMHR) regains ever-increasing attention, mainly benefiting from the excellent representation capability of deep neural networks. On other hand, vulnerability networks exposes a cross-modal retrieval system to various safety risks (e.g., adversarial attack). However, attacking remains underexplored. In this paper, we propose an effective Adversarial Attack on Deep Retrieval, dubbed AACH, which fools target CMHR model in black-box setting....