A5053902963- Digital Media Forensic Detection
- Advanced Steganography and Watermarking Techniques
- Adversarial Robustness in Machine Learning
- Advanced Neural Network Applications
- Anomaly Detection Techniques and Applications
- Advanced Image Processing Techniques
- Generative Adversarial Networks and Image Synthesis
- IoT and Edge/Fog Computing
- Advanced Image Fusion Techniques
- Data Management and Algorithms
- Caching and Content Delivery
- Advanced Data Compression Techniques
- Image and Signal Denoising Methods
- Image Enhancement Techniques
- Biometric Identification and Security
- Chaos-based Image/Signal Encryption
- Advanced Database Systems and Queries
- Energy Efficient Wireless Sensor Networks
- Educational Innovations and Challenges
- Hand Gesture Recognition Systems
- Ideological and Political Education
- Image Processing Techniques and Applications
- Privacy-Preserving Technologies in Data
- Brain Tumor Detection and Classification
- Lung Cancer Diagnosis and Treatment
Yunnan University
2015-2024
Taiyuan University of Technology
2023
Despite achieving exceptional performance, deep neural networks (DNNs) suffer from the harassment caused by adversarial examples, which are produced corrupting clean examples with tiny perturbations. Many powerful defense methods have been presented such as training data augmentation and input reconstruction which, however, usually rely on prior knowledge of targeted models or attacks. A example its version very similar but different high-level representations in a victim model. If we can...
Although several cloud storage systems have been proposed, most of them can provide highly efficient point queries only because the key-value pairs storing mechanism. For these systems, satisfying complex multi-dimensional means scanning whole dataset, which is inefficient. In this paper, we propose a multidimensional index framework, based on Skip-list and Octree, refer to as Skip-Octree. Using randomized skip list makes hierarchical Octree structure easier implement in system. To support...
Abstract Image hiding is a task that hides secret images into cover images. The purposes of image are to ensure the invisible human and can be recovered. current state‐of‐the‐art steganography methods run risk information leakage. A safe network (SIHNet) presented reduce leakage information. Based on some phenomena which use invertible neural network, reversible processing (SIP) module proposed make suitable for stego leak less Besides, lost (LIH) used hide images, thus method recover better...
Abstract Image steganography is the art of concealing secret information within images to prevent detection. In deep‐learning‐based image steganography, a common practice fuse with cover directly generate stego image. However, not all features are equally critical for data hiding, and some insignificant ones may lead appearance residual artifacts in this article, novel network architecture hybrid attention mechanism based on generative adversarial introduced. This model consists three...
Abstract Image hiding is a task that embeds secret images in digital without being detected. The performance of image has been greatly improved by using the invertible neural network. However, current methods are less robust face Joint Photographic Experts Group (JPEG) compression. cannot be extracted from stego after JPEG compression image. Some show good robustness for some certain quality factors but poor other common factors. An image‐hiding network (RIHINNet) to all proposed. First all,...
Unmanned aerial vehicles (UAVs) have gained considerable attention in the research community due to their exceptional agility, maneuverability, and potential applications fields like surveillance, multi-access edge computing (MEC), various other domains. However, efficiently providing computation offloading services for concurrent Internet of Things devices (IOTDs) remains a significant challenge UAVs limited communication capabilities. Consequently, optimizing managing constrained...
Honeycomb lung is a pulmonary manifestation that occurs in the terminal stage of various diseases, which greatly threatens patients. Due to different locations and irregular shapes lesions, accurate segmentation honeycomb region an essential challenging problem. However, most deep learning methods struggle effectively utilize both global local information from lesion images, resulting cannot accurately segment lesion. In addition, these often ignore some semantic necessary for location shape...
Nowadays the information explosion have generated a large amount of data. To utilise these data, there is trend setup efficient data index, especially for multidimensional indexing. However, most current cloud storage systems build indexing based on distributed hash (DHT), where are stored by key-value. This module not very suitable range queries in solve problems, this paper builds index skip lists and octree structure. Firstly, architecture adopts structure to store establish corresponding...
Most cloud providers offer Deep Learning as a Service (DLaaS) for different business, science and engineering domains. However, it is known that deep neural networks (DNNs) are vulnerable to adversarial examples, which can cause well-trained DNN models misbehave by injecting human-imperceptible perturbations the query input data. Securing learning service becomes critical challenge in mitigating such perturbations, enhancing robustness of DNNs. In this article, we report two important facts:...
The Royal Society of Chemistry is the world's leading chemistry community.
Image preprocessing models are usually employed as the preceding operations of high‐level vision tasks to improve performance. The adversarial attack technology makes both these face severe challenges. Prior research is focused solely on attacking single object detection models, without considering impact (multifocus image fusion) perturbations within system. Multifocus fusion work in conjunction with enhance quality images and capability Herein, problem system that utilizes multifocus its...
In the rapidly evolving domain of mobile edge–cloud computing (MECC), proliferation Internet Things (IoT) devices and applications poses significant challenges, particularly in dynamically managing computational demands user mobility. Current research has partially addressed aspects service migration resource allocation, yet it often falls short thoroughly examining nuanced interdependencies between strategies consequential impacts delays, intricacies handling incomplete tasks during...
Abstract Image steganography aims to hide secret data in the cover media for covert communication. Though many deep‐learning‐based image methods have been presented, these approaches suffer from inefficiency of building long‐distance connections between and images, leading noticeable modification traces poor steganalysis resistance. To improve visual imperceptibility generated stego it is essential establish a global correlation images. In this way, can be dispersed throughout globally....
Adversarial examples in which imperceptible perturbations to the input can easily subvert a well-trained model’s prediction pose huge potential security threats deep neural networks (DNNs). As an effective way resist adversarial samples, reconstruction eliminate antagonism of inference process without involving modifications target structure and parameters. However, preprocessing inputs often results some loss protected accuracy. In this paper, we introduce new method that adopts high-level...
With the arrival of era "new engineering", rapid development Internet and new technologies has provided references for teaching models in universities. Traditional can no longer meet needs contemporary college students. Integrating rain classroom, China MOOCs, Bilibili other platforms into process will play an important role "1 + 1> 2". Especially national strategic disciplines such as cyberspace security artificial intelligence, it is necessary to closely align with needs, actively adjust...
Multi-focus image fusion can overcome the issues that optical lens imaging cannot focus multiple targets simultaneously due to depth of field limitation. In this paper, we propose a generative adversarial network (DDF-GAN) which consists generator and two discriminators directly generate fused images without decision maps post-processing. training process, source all-in-focus generated by are used as input one dual discriminators. Meanwhile, gradient map another discriminator. An...
The robustness and security of deep neural network (DNN) models have received much attention in recent years. In-depth research on adversarial example generation methods that make DNN wrong judgments decisions will facilitate further more comprehensive practical defense methods. Most existing focus too attack performance design noise at the pixel level, resulting generated examples with redundant evident perturbations. In this paper, we try to find well-designed perturbations feature-level...