- Advanced Steganography and Watermarking Techniques
- Peer-to-Peer Network Technologies
- Web Data Mining and Analysis
- Digital Media Forensic Detection
- Data Management and Algorithms
- Mobile Agent-Based Network Management
- Rough Sets and Fuzzy Logic
- Advanced Image and Video Retrieval Techniques
- Multimodal Machine Learning Applications
- Machine Learning in Healthcare
- Stock Market Forecasting Methods
- Algorithms and Data Compression
- Distributed and Parallel Computing Systems
- Advanced Computational Techniques and Applications
- Generative Adversarial Networks and Image Synthesis
- Advanced Database Systems and Queries
- Image and Video Stabilization
- Service-Oriented Architecture and Web Services
- Opportunistic and Delay-Tolerant Networks
- RFID technology advancements
- Network Traffic and Congestion Control
- Fuzzy Logic and Control Systems
- Human Pose and Action Recognition
- Internet Traffic Analysis and Secure E-voting
- Data Mining Algorithms and Applications
Guangxi University
2013-2025
Huazhong University of Science and Technology
2008-2009
Capital University of Economics and Business
2009
With the enhancement of information volume, people are not satisfied with transmitting only a single secret image at time but chase to hide multiple images in picture; however, large-capacity steganographic scale can easily lead degradation quality image, which attracts attention eavesdroppers. In this paper, we propose Chaotic mapping-enHanced imAge Steganography nEtwork (CHASE), pioneers colour grey and reduces difference between container cover through permutation method, so as enhance...
With the advancement of technology, information hiding capacity has significantly increased, allowing a cover image to conceal one or more secret images. However, this high often leads contour shadows and color distortions, making high-quality recovery images extremely challenging. Existing algorithms based on Invertible Neural Networks (INNs) discard useful during process, resulting in poor quality recovered images, especially multi-image scenarios. The theoretical symmetry INNs ensures...
Futures commodity prices are affected by many factors, and traditional forecasting methods require close attention from professionals suffer high subjectivity, slowness, low accuracy. In this paper, we propose a new method for predicting the fluctuation in futures accurately. We solve problem of slow convergence ordinary artificial bee colony algorithms introducing population chaotic mapping initialization operator use resulting algorithm as trainer to learn long short-term memory neural...
In Linguistic steganography, when secret information is embedded into to the text using synonym substitution-base method, obvious mistakes and logical misconceptions, resulted from inaccuracy of candidate synonyms, are ubiquitous. A new steganography algorithm proposed based on vector distance two-gram dependency collocations. The can calculate appropriateness substitution. Firstly, which has same attribute semantic as target word chosen WordNet sets. Secondly, collocations should be...
Automatically generating an accurate and meaningful description of image is very challenging. However, the recent scheme caption by maximizing likelihood target sentences lacks capacity recognizing human-object interaction (HOI) semantic relationship between HOIs scenes, which are essential parts caption. This article proposes a novel two-phase framework to generate addressing above challenges: 1) hybrid deep learning 2) generation. In deep-learning phase, factored three-way machine was...
(1) Background: We aimed to develop a transparent machine-learning (ML) framework automatically identify patients with condition from electronic health records (EHRs) via parsimonious set of features. (2) Methods: linked multiple sources EHRs, including 917,496,869 primary care and 40,656,805 secondary 694,954 specialist surgeries between 2002 2012, generate unique dataset. Then, we treated patient identification as problem text classification proposed disease-phenotyping framework. This...
Data mining is an important technique in summarize and prediction different industry. With the help of grid computer, capability for data storage efficiency process can be highly increased. Nowadays, most studies focus on replication mechanism replica selection method, but ignore part service which also important. In paper, we propose a dynamic (DSRP) system based service-oriented architecture (SOA). DSRP has to create delete automatically achieve better load-balancing performance. process,...
The type-1 ordered weighted averaging (T1OWA) operator has demonstrated the capacity for directly aggregating multiple sources of linguistic information modeled by fuzzy sets rather than crisp values. Yager's (OWA) operators possess properties idempotence, monotonicity, compensativeness, and commutativity. This article aims to address whether or not T1OWA these when inputs associated weights are instead numbers. To this end, a partially relation is defined based on maximum (join) minimum...
Traditional distributed file system in availability, scalability, and data access performance has been difficult to meet the growing demand for storage. This paper proposes a Metadata Server (MDS) Cluster scheme based on two-server high availability. On this basis, use combination of directory subtree partitioning improved consistency hash algorithm divides metadata, ensure cluster load balancing adapted change size cluster. For imbalance caused by difference during operation, propose...
P2P network is a great innovation makes computing and data transmission more efficiency convenience. With the rapid development of network, many problems appear to our sight. Nowadays, most studies focus on routing algorithms structured topology but ignore foundation: replication mechanism. In paper, research based sub dividable area. We propose new mechanism calls junction method (JRM) in unstructured decentralized which reflects usual real environment. method, we can achieve better...
Because RFID does not require line of sight communication, low-cost and efficient operation with these outstanding advantages are being widely used, followed by privacy security vulnerabilities other issues. Afterdescribe analysis the facing issues existing protocols on stage, proposed a authentication protocol based hash function, this use function random numbers to ensure safe control access between tags readers,and from perspective quantitative estimates cost tag. After setting up...
Index compression is an effective way to solve the problem that index too huge when mass information retrieved. Based on Streamline Dynamic Successive Trees(SDST) index, a method with querying without decompression proposed in this paper.It also presents algorithm,the algorithm of how retrieve decompression.The experiments show efficiency SDST much better than IF and its time less.
Secure index is the core technology in full-ciphertext retrieval. In order to appliance efficient full-text retrieval ciphertext state, a streamline dynamic successive trees of ciphertext(SDSTC)index model proposed. The SDSTC which supports queries substring and updating index, has high recall precision ratio Giving algorithms creation, analysis its security efficiency. Compared with other existing models, experiments show that good time efficiency more suitable for areas
The effect of the Stream line Inter-relevant Successive Trees (SIRST) is great in field text search. In this paper, binary SIRST will be extended to a ternary one. Whats more, we propose method calculating K K-ary SIRST. Finally, sliding window retrieval algorithm proposed. Experimental results show that more efficient than one mass information retrieval, and search normal
Based on Streamline Dynamic Successive Trees, the paper proposes a dynamic index update method which size is in character-level, The document-level algorithm and character-level are based SDST, performance of these 2 algorithms compared, efficiency proved by experiment. result shows that better than document-level.
Automatically describing the content of an image is a challenging task in computer vision that connects machine learning and natural language processing. In this paper, we present framework, based on modeling context, to generate sentences image, which consists two parts: relation description generating. By mapping from spatial context logical relationship between objects, former trained maximize likelihood target linguistics phrase object given training image. taking advantages...