With the development of smart manufacturing in Industry 4.0, large amount of heterogeneous data are generated from multiple sources. Various dataprocessing techniques can be applied to these data for the purpose of e...
详细信息
It is observed in the literature that data augmentation can significantly mitigate membership inference (MI) attack. However, in this work, we challenge this observation by proposing new MI attacks to utilize the info...
详细信息
Top-k similarity join on high-dimensional data plays an important role in many applications. The traditional tree-like index based approaches can’t deal with large scale high-dimensional data efficiently because of &...
详细信息
Motivation: 3D neuron segmentation is a key step for the neuron digital reconstruction, which is essential for exploring brain circuits and understanding brain functions. However, the fine line-shaped nerve fibers of ...
详细信息
Verifying the correctness of a textual statement requires not only semantic reasoning about the meaning of words, but also symbolic reasoning about logical operations like count, superlative, aggregation, etc. In this...
详细信息
In this paper,we study the dynamical properties of Gstrong chain recurrent point,G-chain point set and G-chain equivalent point of topological G-conjugacy on metric *** inference,we give the following conclusions that...
详细信息
ISBN:
(纸本)9781728150505
In this paper,we study the dynamical properties of Gstrong chain recurrent point,G-chain point set and G-chain equivalent point of topological G-conjugacy on metric *** inference,we give the following conclusions that if let f:X→X and f:Y→Y be two continous map of metric G-space X and *** the map h:X→Y is a topogical G-conjugacy from f to f,then（1） h(SCR（f）)=SCR（f）;（2） h(S（x,f）)=S(h（x）,f);（3）h(CE（x,f）)=CE(h（x）,f).These results will enrich the theory of G-strong chain recurrent point,G-chain point and G-chain equivalent point of topological G-conjugacy on metric G-space
In scenarios where multiple parties such as the Internet of Things and Supply Chains participate in data sharing and computing, when accessing data, users not only need to accept the forward access control of the data...
详细信息
In scenarios where multiple parties such as the Internet of Things and Supply Chains participate in data sharing and computing, when accessing data, users not only need to accept the forward access control of the data owner, but also needs to perform reverse access control on the data, so as to realize two-way access control. In the traditional blockchain, all users can participate in accounting and view transaction data, and only protect user privacy through “pseudo-anonymity”. The access rights of different users cannot be distinguished, which cannot meet the user's two-way access control needs. However, most of the existing blockchain-based access control schemes are one-way access control, which cannot meet the needs of users for two-way access control in scenarios such as the Internet of Things and Supply Chains. Therefore, it is particularly important to design a two-way access control mechanism suitable for application in the blockchain. On this basis, this paper proposes a dual strategy attribute-based encryption (ABE) scheme for distributed outsourcing. This scheme combines two existing schemes, ciphertext-policy ABE and key-policy ABE, and proposes two access structures and a structure of attribute sets. The primary access structure and the secondary attributes are stored in the ciphertext, and the secondary access structure and the primary attributes are stored in the user's private key. Only when the primary attribute set satisfies the primary access structure and the secondary attribute set satisfies the secondary access structure can the user unlock the ciphertext. This scheme has no single authorization center; instead, blockchain nodes jointly participate in authorization. In addition, the proposed scheme outsources the encryption and decryption of the ciphertext to blockchain nodes to reduce the computing pressure on users and can adapt to the decentralized environment of the blockchain and provide users with two-way access control services. Finally
image captioning is usually based on the mainstream encoder-decoder framework, the accuracy of the encoders and the generation capability of the decoders directly affect the quality of image captions. But no matter ho...
详细信息
ISBN:
(纸本)9781450384087
image captioning is usually based on the mainstream encoder-decoder framework, the accuracy of the encoders and the generation capability of the decoders directly affect the quality of image captions. But no matter how the encoders and the decoders are combined, the generated captions will only imitate the ground-truth captions, and its diversity is suppressed. To generate more accurate and diverse captions while exploring the impact of different encoder-decoder frameworks for the quality of the generated captions, we conducted an exploration of an efficient image captioning method based on GAN. Specifically, we use ResNet152, EfficientNet-B0, and EfficientNet-B2 as the encoders to extract image features, GRU or Transformer as the decoders to generate captions, respectively. Based on MS COCO2014 dataset, a large number of experiment results show that the EfficientNet-B2+GRU framework has best performance, training time is reduced, BLUE-4 score increases by 0.6, parameters reduce by 0.66 million, which can be used as an efficient encoder-decoder framework for generating diverse captions.
Three issues reduced the performance of networks for handling the organs and lesions’ simultaneous segmentation in electronic laryngoscopy images. Firstly, the moving endoscope will cause noticeable variations of the...
详细信息
The major work is to retrieve the surface soil moisture under crop cover in this paper. This paper proposes an improved GA-BP model called as DEFS-GA-BP model. This model is proposed to retrieve the surface soil moist...
详细信息
The major work is to retrieve the surface soil moisture under crop cover in this paper. This paper proposes an improved GA-BP model called as DEFS-GA-BP model. This model is proposed to retrieve the surface soil moisture distribution using Sentinel-1 and Sentinel-2 remote sensing data. After extracting nine polarization features and one roughness parameter from Sentinel-1 data, five vegetation indexes from Sentinel-2 data. The differential evolution feature selection (DEFS) algorithm is applied to select the optimal feature subset. By using the genetic algorithm (GA) to optimize the node weights, the back propagation (BP) neural network is trained using the optimal feature subset and then used to generate the surface soil moisture distribution. The results of inversion algorithm indicate that the proposed DEFS-GA-BP model has a better inversion precision than GA-BP model, and demonstrates its application potential in surface soil moisture inversion.
暂无评论