The task of information platform is to manage, gather and share data in correlative departments. The function of information platform is in two ways. First, it is the information hinge across the correlative departmen...
详细信息
Addressing some vital security issues in enterprise grid technologies is particularly important from the software industry's viewpoint. Enterprise grid computing systems do face unique security challenges, ranging...
详细信息
This paper proposes a hyper-converged blockchain hardware and software sys-tem based on domestic chips. By utilizing domestically firmware, the Linux operating system, X86 architecture Hygon processors, and ARM archit...
详细信息
The use of online social networks (OSN) has been a reality for some years now. If OSN users have mobile devices and they are taken into account, every day a vast amount of information is spontaneously generated, creat...
详细信息
E-commerce web sites usually have to maintain a large number of product information. To organize this product information, a feasible way is to add semantic tags to the information. However, the Web product informatio...
详细信息
Systematic level design and optimization of a Radio Frequency (RF) receiver is a multidimensional problem and it is usually an arduous and experience-based work because of the trade-off between various system paramete...
详细信息
ISBN:
(纸本)9783037853801
Systematic level design and optimization of a Radio Frequency (RF) receiver is a multidimensional problem and it is usually an arduous and experience-based work because of the trade-off between various system parameters. In the mean while, co-operated with Genetic Algorithm (GA) and local search, Memetic Algorithm (MA) gradually becomes popular to solve multidimensional optimization problems, and it can be an efficient method for computer-aided design of RF receivers. In this paper, the MA method was adopted to aid the design of a 5.8 GHz RF receiver front-end. The transfer function for the signal-to-noise ratio at the front-end output (SNRout) was first derived based on the gain, noise figure and inter-modulation product of each individual circuit component, and the transfer function was then used as the required fitness function for the MA method. The efficacy of the proposed method was validated by ADS simulations.
With the recent spreading of social networks and digital communities, the Web has a new business model for providing free content-sharing services. Using the power of the masses and the collective intelligence, Web 2....
详细信息
ISBN:
(纸本)9781601321480
With the recent spreading of social networks and digital communities, the Web has a new business model for providing free content-sharing services. Using the power of the masses and the collective intelligence, Web 2.0 sites have created enormous e-commerce opportunities for marketing and advertising. Thus, there is still a necessity of developing tools to help users to reach good decisions in such frameworks. We propose to manage this collective intelligence with a Decision Support System (DSS). It allows that users with similar user profiles that the potential customer can advice him/her according with their own experiences in their e-commerce activities. From user preferences the DSS obtains a collective advice that helps customers to make decisions related with their e-commerce activities.
This study investigates an approach that combines text and image data to classify web pages based on their content. In this methodology, Convolutional Neural Networks (CNN) are used to analyze text data, while the YOL...
详细信息
Federated Learning (FL) is a paradigm that allows users to collaboratively train machine learning models by sharing gradients during the training process without sharing the original data. In practical application sce...
详细信息
Federated Learning (FL) is a paradigm that allows users to collaboratively train machine learning models by sharing gradients during the training process without sharing the original data. In practical application scenarios, the server acts as a coordinator, collecting and aggregating user gradients. Driven by commercial interests, the server may be inclined to provide users with incorrect aggregated results. Nevertheless, only a limited number of research schemes have focused on developing a verifiable FL framework, these schemes have not yet been able to resist collusion behaviors effectively. In this paper, we meticulously examine the verification mechanism of VCD-FL [1] and propose a model recovery attack and a forgery attack, demonstrating that the previous verification mechanism has privacy leakage issues. Subsequently, we propose a framework named VCFL, which aims to achieve verifiable, collusion-resistant, and communication-efficient FL. VCFL introduces an optimized double-masking algorithm, and an Elliptic Curve Cryptography (ECC)-based commitment scheme, thereby protecting the privacy of gradients during the FL process and providing effective verification of aggregated results. Experimental results demonstrate that VCFL is 60% faster in model training speed and reduces the communication overhead on the user side by 17%, as compared to VerifyNet. Additionally, the verification overhead of VCFL is independent of the dimension.
With technologies that have democratized the production and reproduction of information, a significant portion of daily interacted posts in social media has been infected by rumors. Despite the extensive research on r...
详细信息
暂无评论