Article
Chemistry, Analytical
Ting Wang, Geng Wei, Huayu Li, Thioanh Bui, Qian Zeng, Ruliang Wang
Summary: This paper proposes a method that combines convolutional neural networks with joint texture recognition to reduce encoding complexity in high-efficiency video coding. By using techniques such as early termination and predictive partitioning, the method achieves a balance between reducing computational complexity and maintaining coding performance.
Article
Multidisciplinary Sciences
Guowei Teng, Danqi Xiong, Ran Ma, Ping An
Summary: In this paper, a decision tree accelerated coding tree units (CTU) partition algorithm is proposed for intra prediction in VVC, which can significantly reduce encoding time with only a slight increase in bit rate compared to the reference test model.
Article
Computer Science, Information Systems
Shuqian He, Zhengjie Deng, Chun Shi
Summary: The proposed method in this paper utilizes a fuzzy support vector machine classifier with information entropy measure to address data noise and outliers issues in HEVC encoding, leading to improved coding performance. Experimental results demonstrate a significant reduction in encoding time for various test video sequences compared to existing algorithms, while maintaining distortion cost performance.
Article
Computer Science, Information Systems
Junaid Tariq, Ammar Armghan, Fayadh Alenezi, Amir Ijaz, Saad Rehman, Ayman Alfalou, Junaid Ali Khan
Summary: This article introduces a fast intra-mode estimation algorithm based on the 'German Tanks Problem', which reduces the compression time of HEVC significantly.
Article
Computer Science, Information Systems
Junaid Tariq, Ayman Alfalou, Amir Ijaz, Hashim Ali, Imran Ashraf, Hameedur Rahman, Ammar Armghan, Inzamam Mashood, Saad Rehman
Summary: Comprehension algorithms like High Efficiency Video Coding (HEVC) enable fast and efficient handling of multimedia contents. However, the brute-force behavior of HEVC poses a challenge in multimedia content transmission. This article presents a novel method to accelerate the encoding process of HEVC by making early intra mode decisions for each block. The proposed method effectively utilizes neighboring blocks to extract information and calculate the best intra mode based on probability rules and optimal stopping theory. The proposed algorithms achieve a 30.22% speedup in the HEVC encoding process with a 1.35% reduction in Bit Rate (BD-BR).
CMC-COMPUTERS MATERIALS & CONTINUA
(2022)
Article
Computer Science, Software Engineering
Qiuwen Zhang, Tengyao Cui, Rijian Su
Summary: The new generation video coding standard H.266/VVC inherits and develops from H.265/HEVC, introducing flexible coding tools and partition structure, and proposes a fast CU partition algorithm based on texture. Experimental results show that the algorithm can achieve an average time saving of 50.56% compared to the anchor algorithm, but with an average increase of 1.31% in BDBR.
SCIENTIFIC PROGRAMMING
(2021)
Article
Computer Science, Artificial Intelligence
Zhewen Sun, Li Yu, Wei Peng
Summary: This paper proposes a fast intra partition algorithm using Lightweight Neural Network (LNN) to skip QTMT partition steps, improving coding efficiency and reducing encoding time specifically for 360-degree videos in the newest video coding standard VVC.
IET IMAGE PROCESSING
(2023)
Article
Engineering, Multidisciplinary
Farid Zaki, Amr E. Mohamed, Samir G. Sayed
Summary: This paper introduces a framework called CtuNet, which uses deep learning techniques to predict the CTU partitions in HEVC standard, reducing computational complexity and maintaining near-optimal results.
AIN SHAMS ENGINEERING JOURNAL
(2021)
Article
Computer Science, Information Systems
Xiwu Shang, Guoping Li, Xiaoli Zhao, Hua Han, Yifan Zuo
Summary: Recently, a new video coding standard called Versatile Video Coding (VVC) has been introduced, which uses quadtree with nested multi-type tree structure to improve coding efficiency. However, the flexible block partition structure leads to a significant increase in coding complexity. To address this issue, a fast CU size decision algorithm is proposed, which utilizes coding and texture information to speed up the intra coding process. Experimental results show that the proposed algorithm can reduce coding complexity by an average of 51.89% with minimal loss of coding performance, and achieve additional time saving compared to state-of-the-art methods.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Werda Imen, Maraoui Amna, Belghith Fatma, Sayadi Fatma Ezahra, Nouri Masmoudi
Summary: The article introduces an important module in the High-Efficiency Video Coding (HEVC) standard, which is the unit partition structure that expands the coding unit shapes and achieves significant compression performance improvement. To speed up the encoding process, a Convolutional Neural Network-based method is proposed to replace the traditional brute force search. The experimental results demonstrate that the proposed method can significantly reduce the encoding time.
SIGNAL IMAGE AND VIDEO PROCESSING
(2022)
Article
Computer Science, Information Systems
Yanjun Wang, Pu Dai, Jinchao Zhao, Qiuwen Zhang
Summary: The paper introduces the versatile video coding standard (VVC) and its improvement in coding performance, as well as the increase in computational complexity. The author proposes a stage grid map and a multi-stage early termination convolutional neural network (MET-CNN) model, and a fast CU partition decision algorithm based on MET-CNN for VVC intra coding optimization. Experimental results show that the proposed scheme significantly reduces the encoding time.
Article
Computer Science, Information Systems
Tong Wu, Shiyi Liu, Feng Wang, Zhenyu Wang, Rongjie Wang, Ronggang Wang
Summary: AVS3, a new video coding standard, introduces a novel early termination mechanism and fast CU partition algorithm, improving coding efficiency and saving significant encoding time.
Article
Computer Science, Information Systems
Jindou Liu, Zhaohong Li, Xinghao Jiang, Zhenzhen Zhang
Summary: This paper proposes a novel multilevel steganography algorithm based on diamond-encoded prediction unit partition modes. The most outstanding contribution of this paper is the introduction of convolutional neural networks (CNNs) to improve visual quality and reduce steganographic video bitrate increases. Experimental results show that the proposed algorithm achieves higher embedding capacity, excellent visual quality, and strong resistance to steganalysis.
IEEE TRANSACTIONS ON MULTIMEDIA
(2022)
Article
Computer Science, Information Systems
Jinchao Zhao, Aobo Wu, Qiuwen Zhang
Summary: This paper proposes a fast CU partition decision algorithm based on support vector machine to reduce the coding complexity for VVC. The algorithm analyzes the proportion of split modes with different CU sizes, selects more reliable correlation features, and designs and trains two SVM models, achieving better coding results compared to VTM7.0.
Article
Computer Science, Information Systems
Maraoui Amna, Werda Imen, Bouaafia Soulef, Sayadi Fatma Ezahra
Summary: The use of machine learning techniques to reduce the complexity of recent video coding standards such as HEVC has garnered attention. This paper investigates the implications of using FSVM and CNN models for partitioning in HEVC intra-prediction, with experimental results showing that the deep CNN approach outperforms the online FSVM approach in complexity reduction.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Dujuan Gu, Jinhe Su, Yibo Xue, Dongsheng Wang, Jun Li, Ze Luo, Baoping Yan
COMPUTER COMMUNICATIONS
(2020)
Article
Automation & Control Systems
Yi Yang, Di Tang, Dongsheng Wang, Wenjie Song, Junbo Wang, Mengyin Fu
ROBOTICS AND AUTONOMOUS SYSTEMS
(2020)
Article
Optics
Min Yang, Pingxue Li, Shun Li, Wenhao Xiong, Kaixuan Wang, Chuanfei Yao, Dongsheng Wang
Summary: This all-fiber amplifier utilizes a passively mode-locked fiber oscillator and an 8 km single-mode fiber stretcher to deliver nanosecond pulses at an ultra-high repetition rate.
Article
Environmental Sciences
Chunxu Wu, Lanfeng Li, Hao Zhou, Jing Ai, Hongtao Zhang, Jialin Tao, Dongsheng Wang, Weijun Zhang
Summary: This study investigated the adsorption performance of sludge-based activated carbon (SBC) on dissolved organic matters (DOMs) removal from sewage, and the modification effect of different types of chemicals on the structure of synthesized SBC. Chemical activation significantly improved the adsorption capacity of MSBC on humic acids (HA) and aromatic proteins (APN), showcasing the importance of surface functional groups on the adsorption capacities of MSBC towards DOMs removal in sewage. Additionally, the study examined the residual molecular weight of DOMs in sewage, showing varying effectiveness of different chemical modifications on different organic matter weights.
JOURNAL OF ENVIRONMENTAL SCIENCES
(2021)
Article
Computer Science, Hardware & Architecture
Pengfei Qiu, Dongsheng Wang, Yongqiang Lyu, Ruidong Tian, Chunlu Wang, Gang Qu
Summary: The study introduces an attack method to break SGX by inducing voltage-oriented hardware faults, which can be controlled completely by software without requiring any security vulnerabilities in the software. By providing transient low voltage to the processor through a controller module and injecting transient faults into the program running in the enclave, the attack is successfully executed to steal the key.
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
(2021)
Article
Engineering, Electrical & Electronic
Shangming Cai, Dongsheng Wang, Haixia Wang, Yongqiang Lyu, Guangquan Xu, Xi Zheng, Athanasios V. Vasilakos
Summary: Edge deep learning is an emerging topic where edge devices collaboratively train a shared model. However, distributed training over edge networks is time-consuming due to transmission procedures. To address this, we propose DynaComm, a scheduler that optimizes communication and computation overlap for efficient training at the network edge.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2022)
Article
Computer Science, Information Systems
Dongsheng Wang, Hongjie Fan, Junfei Liu
Summary: The study introduces a cross-document NER model that improves entity recognition accuracy by establishing internal relationships and calculating cross-document representations. By adding a multi-classification auxiliary task and employing multi-objective optimization, the model performance and effectiveness are enhanced.
INFORMATION SCIENCES
(2021)
Article
Chemistry, Physical
Daxin Zhang, Yili Wang, Junyi Li, Xiaoyang Fan, Enrui Li, Shuoxun Dong, Weiwen Yin, Dongsheng Wang, Baoyou Shi
Summary: This study proposed an electrical impedance spectroscopy (EIS) method and constructed a generalized framework to associate macroscale electrical properties with microscopic structure and size-related characteristics of aggregates. The models extracted via EIS were capable of describing the self similarity of aggregates and capturing the fractal and size information. The EIS method exhibited a wide range of applications in water and wastewater treatment.
JOURNAL OF COLLOID AND INTERFACE SCIENCE
(2022)
Article
Mathematical & Computational Biology
Zhen Li, Tong Li, YuMei Wu, Liu Yang, Hong Miao, DongSheng Wang
Summary: This paper implements software defect prediction based on deep learning, combining the particle swarm algorithm and the wolf swarm algorithm for optimization, resulting in better performance indicators. By using a hybrid algorithm in the search for model hyperparameter optimization, the model's performance is enhanced.
COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE
(2021)
Article
Computer Science, Information Systems
Chengzhi Wang, Tianqi Yang, Min Han, Dongsheng Wang
Summary: Recently, hybrid logic circuits based on magnetic tunnel junctions (MTJs) have been investigated to reduce standby power. However, these circuits face reliability issues due to limited TMR ratio of the MTJ and process variation in deep sub-micrometer technology node. This paper proposes a novel differential sensing amplifier (DSA) that achieves a large sensing margin by incorporating two PMOS transistors and demonstrates its functionality and performance through simulations.
Proceedings Paper
Computer Science, Hardware & Architecture
Tongliang Li, Haixia Wang, Airan Shao, Dongsheng Wang
Summary: This paper discusses the challenges faced by Persistent Memory (PM) B+-tree designs and introduces a new variant called Side-to-Side B+Tree (SSB-Tree) that utilizes Lazy-Box technology to reduce consistency cost and achieve efficient concurrency protocol and instant recovery with a single 8-byte write. The experimental results show that SSB-Tree outperforms other state-of-the-art PM B+-trees in terms of throughput in various benchmark tests.
2022 IEEE 36TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS 2022)
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Xia'nan Zhao, Dongsheng Wang
Summary: In this paper, a new identity privacy preserving public auditing scheme is proposed to ensure the integrity of group data in mobile cloud storage services. By introducing a third party public auditor and utilizing the Diffie-Hellman Key Exchange protocol, the scheme supports efficient group dynamics while protecting user privacy.
2022 IEEE/ACM 7TH SYMPOSIUM ON EDGE COMPUTING (SEC 2022)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Liyuan Cao, Yingwen Chen, Kaiyu Cai, Dongsheng Wang, Yuchuan Luo, Guangtao Xue
Summary: In the smart grid, smart meters record and transmit real-time electricity consumption data to the control center. To protect user privacy, researchers propose privacy-preserving data aggregation schemes based on aggregate values instead of raw data. However, these schemes overlook the problem of collusion. This paper proposes a collusion-tolerant and privacy-preserving data aggregation scheme for the smart grid, which effectively protects the privacy of customer usage data even under collusions.
WIRELESS ALGORITHMS, SYSTEMS, AND APPLICATIONS (WASA 2022), PT I
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Dongsheng Wang, Hongjie Fan, Junfei Liu
Summary: This paper introduces a new approach for DDI extraction using sequence features, dependency characteristics, capsule network, and an improved loss function, which effectively enhances the performance of DDI extraction.
DATABASE SYSTEMS FOR ADVANCED APPLICATIONS (DASFAA 2021), PT II
(2021)
Article
Computer Science, Information Systems
Liu Yang, Zhen Li, Dongsheng Wang, Hong Miao, Zhaobin Wang
Summary: The paper focuses on software quality, software failure prediction, and software reliability model parameter estimation, proposing a hybrid algorithm (SSA-PSO) for software defect prediction. Experimental results show that the hybrid algorithm has faster convergence speed, more stable, accurate results, solving the issues present in traditional single algorithms.