A5100376927- Autonomous Vehicle Technology and Safety
- Topic Modeling
- Wireless Signal Modulation Classification
- Indoor and Outdoor Localization Technologies
- Natural Language Processing Techniques
- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Traffic control and management
- Speech Recognition and Synthesis
- Traffic Prediction and Management Techniques
- Speech and Audio Processing
- Power Systems and Renewable Energy
- Radar Systems and Signal Processing
- Advanced Neural Network Applications
- Privacy-Preserving Technologies in Data
- Video Surveillance and Tracking Methods
- Vehicular Ad Hoc Networks (VANETs)
- Music and Audio Processing
- Machine Learning and ELM
- Sentiment Analysis and Opinion Mining
- Power Line Communications and Noise
- Robotic Path Planning Algorithms
- Time Series Analysis and Forecasting
- Advanced MIMO Systems Optimization
- Data Quality and Management
Anhui University
2025
National University of Defense Technology
2021-2025
China Centre for Resources Satellite Data and Application
2024
Ministry of Natural Resources
2024
Civil Aviation University of China
2022-2024
Beihang University
2019-2024
Inner Mongolia University of Technology
2024
Beijing University of Posts and Telecommunications
1999-2024
Nanjing Normal University
2016-2024
Hunan University
2024
A central challenge in training classification models the real-world federated system is learning with non-IID data. To cope this, most of existing works involve enforcing regularization local optimization or improving model aggregation scheme at server. Other also share public datasets synthesized samples to supplement under-represented classes introduce a certain level personalization. Though effective, they lack deep understanding how data heterogeneity affects each layer model. In this...
We propose HoWiES, a system that saves energy consumed by WiFi interfaces in mobile devices with the assistance of ZigBee radios. The core component HoWiES is WiFiZigBee message delivery scheme enables radios to convey different messages devices. Based on WiFi-ZigBee scheme, we design three protocols target at saving opportunities scanning, standby and wakeup respectively. have implemented two platforms AP platforms. Our real-world experimental evaluation shows our can thousands from an...
This paper describes the Arabic Multi-Genre Broadcast (MGB-2) Challenge for SLT-2016. Unlike last year's English MGB Challenge, which focused on recognition of diverse TV genres, this year, challenge has an emphasis handling diversity in dialect speech. Audio data comes from 19 distinct programmes Aljazeera channel between March 2005 and December 2015. Programmes are split into three groups: conversations, interviews, reports. A total 1,200 hours have been released with lightly supervised...
Discretionary lane change (DLC) is a basic but complex maneuver in driving, which aims at reaching faster speed or better driving conditions, e.g., further line of sight ride quality. Although modeling DLC decision-making has been studied for years, the impact human factors, crucial accurately modelling strategies, largely ignored existing literature. In this paper, we integrate factors that are represented by styles to design new model. Specifically, our proposed model takes not only...
Inference of detailed vehicle trajectories is crucial for applications such as traffic flow modeling, energy consumption estimation, and optimization. Static sensors can provide only aggregated information, posing challenges in reconstructing individual trajectories. Shockwave theory used to reproduce oscillations that occur between sensors. However, the emerging connected vehicles grows, probe data offers significant opportunities more precise trajectory reconstruction. Existing methods...
Massive trajectory data stems from the prevalence of equipment supporting GPS and wireless communication technology. Based on these data, computation similarity has become a research hotspot in spatial database during recent years. The sampling problem caused by different strategies device many negative effects measurement. Although studies have solved this complements, methods still drawbacks because only temporal features are considered. Activity trajectory, with development LBSN...
Despite the recent advancements in language models (LMs), their ability to solve complex problems remains limited. This paper introduces Cumulative Reasoning (CR), a novel approach that utilizes LMs cumulatively and iteratively, mirroring human thought processes for problem-solving. CR decomposes tasks into smaller, manageable components leverages previous propositions effective composition, significantly enhancing problem-solving capabilities. We demonstrate CR's superiority through several...
We present Fanar, a platform for Arabic-centric multimodal generative AI systems, that supports language, speech and image generation tasks. At the heart of Fanar are Star Prime, two highly capable Arabic Large Language Models (LLMs) best in class on well established benchmarks similar sized models. is 7B (billion) parameter model was trained from scratch nearly 1 trillion clean deduplicated Arabic, English Code tokens. Prime 9B continually Gemma-2 base same token set. Both models...
Web applications have become a crucial part of modern society’s infrastructure, and vulnerabilities in them can lead to significant social economic losses. Static analysis remains the predominant approach for vulnerability detection, due its extensive coverage. However, high false positive rate demands expert effort confirm actual presence vulnerabilities. In contrast, dynamic generate accurate reports. Nevertheless, existing fuzzers are often constrained their methodologies, making it...
This paper describes a detailed comparison of several state-ofthe-art speech recognition techniques applied to limited Arabic broadcast news dataset. The different approaches were all trained on 50 hours transcribed audio from the Al-Jazeera channel. best results obtained using i-vectorbased speaker adaptation in training scenario Minimum Phone Error (MPE) criteria combined with sequential Deep Neural Network (DNN) training. We report for two types test data: reports, word error rate (WER)...
Abstract A mixed strategy improved dung beetle optimization (MSDBO) algorithm is proposed to address the problems of slow convergence speed, easy falling into local optimum, and insufficient search accuracy algorithm. Firstly, good point set introduced initialize population improve diversity. Then, spiral combined with whale location update reproduction foraging behavior, balancing exploitation global ability algorithm, improving Finally, Levy flight used stealing behavior algorithm's jump...
Transportation systems relying on vehicles to collect data for services such as road condition monitoring are vulnerable malicious injecting large amounts of fake data. A particularly serious type attack is one in which attackers report information about numerous places without actually having been there. In this paper, we present VProof, a vehicle location proof scheme that enables prove its claims match historical locations. With construct their proofs by simply extracting relevant...
Sampling-based motion planning (SBMP) is a major trajectory approach in autonomous driving given its high efficiency practice. As the core of SBMP schemes, sampling strategy holds key to whether smooth and collision-free can be found real-time. Although some bias strategies have been explored literature accelerate SBMP, generated under existing may lead sharp lane changing. To address this issue, we propose new learning framework for SBMP. Specifically, develop novel automatic labeling...
For autonomous driving, an essential task is to detect surrounding objects accurately. To this end, most existing systems use optical devices, including cameras and light detection ranging (LiDAR) sensors, collect environment data in real time. In recent years, many researchers have developed advanced machine learning models objects. Nevertheless, the aforementioned devices are vulnerable signal attacks, which could compromise accuracy of object detection. address critical issue, we propose...
Barrier certificates provide safety guarantees for hybrid systems. In this paper, we propose a novel approach to synthesizing neural networks as barrier certificates. Candidate are trained from special structure: ReLU consisting of two hidden layers. Then, the problem identifying real candidates is transformed into group mixed integer linear programming problems and quadratically constrained problem. Taking full advantage recent advance in optimization, validation can be performed...
Sampling-based motion planning (SBMP) is a major algorithmic trajectory approach in autonomous driving given its high efficiency and outstanding performance practice. However, safety still calls for further refinement of SBMP. In this article we organically integrate with learning models to improve SBMP highway traffic scenarios from the following two perspectives. First, number points be sampled, develop new model sample “important” by predicting intention surrounding vehicles distribution...
The Internet of Things (IoT) permeates every aspect our daily lives as billions interconnected devices are deployed in the physical world. However, IoT networks operate an untrusted environment and often suffer from many malicious active attacks. Automatic modulation classification (AMC), which can identify format intercepted signals without prior knowledge, is a vital technology countering physical-layer threats IoT. most existing algorithms assume channel time invariant, AMC time-varying...
As a new type of flexible regulation resource, energy storage systems not only smooth out the fluctuation generation but also track scheduling combined with power to enhance reliability system operations. In recent years, installing for on-grid stations has become basic requirement in China, there is still lack relevant assessment strategies and techno-economic evaluation size determination from perspective stations. Therefore, this paper starts summarizing role configuration method then...
With the population ages, many patients are unable to receive comprehensive care, leading an increase in hazardous incidents, particularly falls occurring after getting out of bed. To address this issue, paper proposes a method for recognizing bed-exit intentions using array air spring mattress. The integrates convolutional neural networks with feature point matching techniques identify both global and local features spring. For features, one-dimensional focal loss network (1D-FLCNN) model...