A broadband methodology for measurement of uniaxial dielectric anisotropy is described. Previous work is improved with reduced measurement error, enhanced automation, and additional material testing. Additionally, per...
详细信息
A broadband methodology for measurement of uniaxial dielectric anisotropy is described. Previous work is improved with reduced measurement error, enhanced automation, and additional material testing. Additionally, permittivity variation with temperature is explored.
In order to explore the efficient design and implementation of activation function in deep learning processor, this paper presents an efficient five-stage pipelined hardware architecture for activation function based ...
详细信息
ISBN:
(纸本)9781538649923
In order to explore the efficient design and implementation of activation function in deep learning processor, this paper presents an efficient five-stage pipelined hardware architecture for activation function based on the piecewise linear interpolation, and a novel neuron data-LUT address mapping algorithm. Compared with the previous designs based on serial calculation, the proposed hardware architecture can achieve at least 3 times of acceleration. Four commonly used activation functions are designed based on the proposed hardware architecture, which is implemented on the XC6VLX240T of Xilinx. The LeNet-5 and AlexNet are selected as benchmarks to test the inference accuracy of different activation functions with different piecewise numbers on the MNIST and CIFAR-10 test sets in the deep learning processor prototype system. The experiment results show that the proposed hardware architecture can effectively accomplish the relevant calculation of activation functions in the deep learning processor and the accuracy loss is negligible. The proposed hardware architecture is adaptable for numerous activation functions, which can be widely used in the design of other deep learning processors.
The paper mainly solved the problem of the design of command and control structure. First, the key elements in command and control structure are defined and we introduced the concept of work load of decision makers to...
详细信息
The paper mainly solved the problem of the design of command and control structure. First, the key elements in command and control structure are defined and we introduced the concept of work load of decision makers to describe the problem mathematically. Next, a mathematic model aimed at minimizing the root mean square of decision makers' work load is developed. Finally, we combine the quantum genetic algorithm with self-adaption strategy and get the self-adaption quantum genetic algorithm. Major characteristic of this algorithm is adjusting the quantum rotation gate, generating the crossover probability and the mutation probability in a self-adaption way. Experimental results show that the self-adaption quantum genetic algorithm has a feature of evolving fast and searching precise, and it can cluster the platforms well to accomplish the design of command and control structure.
Speed and cost of logistics are two major concerns to on-line shoppers, but they generally conflict with each other in nature. To alleviate the contradiction, we propose to exploit existing taxis that are transporting...
详细信息
Magic square construction is a complex and hard permutation problem of recreational combinatorics with a long history. Not only does the complexity consist in that the number of magic squares increases rapidly with th...
详细信息
ISBN:
(纸本)0780378040
Magic square construction is a complex and hard permutation problem of recreational combinatorics with a long history. Not only does the complexity consist in that the number of magic squares increases rapidly with the order of magic square, but also in that the percentage of magic squares in the possible permutation of the first n/sup 2/ natural numbers decreases with the order. Based on the two-phase construction conjecture, we propose an improved evolutionary algorithm for magic square construction. Mutation operators are specially designed so that the mutation domain can be located and the mutation probabilities adjusted adaptively, which include the number pair permutations, the row permutations and column permutations. In addition, some heuristics-based local permutations, like the row/column local rectification and the local diagonal rectification, are used to complement the stochastic mechanism. Computational results show that, the two-phase construction conjecture is computationally effective, and the improved evolutionary algorithm is highly efficient for magic square construction.
Generative adversarial network (GAN) has shown its powerful capability in generating photorealistic images. Although the generated images can fool human eyes, it is not clear whether they can evade the detection of fo...
详细信息
ISBN:
(纸本)9781728102436;9789881476852
Generative adversarial network (GAN) has shown its powerful capability in generating photorealistic images. Although the generated images can fool human eyes, it is not clear whether they can evade the detection of forensic detectors, which aim to identify the originality and authenticity of images. In this paper, we investigate how forensic detectors perform in differentiating between GAN generated images and real images. We consider two kinds of approaches, one is intrusive and the other is non-intrusive, based on whether the GAN architecture is needed for performing detection. We have conducted extensive experiments on a celebrity face image dataset to evaluate the effectiveness of different approaches. The results and analyses show that the intrusive approach can detect GAN generated images but with a relatively high false alarm rate. The non-intrusive approach with features extracted from a VGG network is very effective for detecting GAN generated images when the training data is sufficient, but it still faces challenge when the training data and testing data are mismatched.
With the advancement of artificial intelligence technology, intelligent applications have permeated all aspects of human life, including automatic driving, intelligent security, and face recognition. Convolutional neu...
详细信息
ISBN:
(数字)9798350387384
ISBN:
(纸本)9798350387391
With the advancement of artificial intelligence technology, intelligent applications have permeated all aspects of human life, including automatic driving, intelligent security, and face recognition. Convolutional neural networks (CNNs) are essential models for AI, but their complexity increases with more available data. The complex models are unsuitable for edge devices like wearables and smartphones due to high memory and computational requirements. In this paper, the compact model is achieved by pruning some redundant output channels. To effectively measure the validity of the output channels, this paper proposes a channel pruning algorithm based on gradient integration, which combines the gradient of the channel with the model output and the channel’s values. This pruning could destroy model accuracy, so knowledge distillation is introduced during fine-tuning to leverage the original model’s reasoning ability. Experiments show that our approach improves accuracy by 0.51% while reducing parameters by 30.1% and FLOPs by 26% in the VGG16 model.
In a wide range of application areas (e.g. data mining, approximate query evaluation, histogram construction), database sampling has proved to be a powerful technique. It is generally used when the computational cost ...
详细信息
In a wide range of application areas (e.g. data mining, approximate query evaluation, histogram construction), database sampling has proved to be a powerful technique. It is generally used when the computational cost of processing large amounts of information is extremely high, and a faster response with a lower level of accuracy for the results is preferred. Previous sampling techniques achieve this balance, however, an evaluation of the cost of the database sampling process should be considered. We argue that the performance of current relational database sampling techniques that maintain the data integrity of the sample database is low and a faster strategy needs to be devised. In this paper we propose a very fast sampling method that maintains the referential integrity of the sample database intact. The sampling method targets the production environment of a system under development, that generally consists of large amounts of data computationally costly to analyze. We evaluate our method in comparison with previous database sampling approaches and show that our method produces a sample database at least 300 times faster and with a maximum trade off of 0.5% in terms of sample size error.
Recently, the application of transfer learning within dynamic multiobjective evolutionary algorithms (DMOEAs) has shown significant potential to solve dynamic multiobjective optimization problems (DMOPs). This approac...
详细信息
ISBN:
(数字)9798350308365
ISBN:
(纸本)9798350308372
Recently, the application of transfer learning within dynamic multiobjective evolutionary algorithms (DMOEAs) has shown significant potential to solve dynamic multiobjective optimization problems (DMOPs). This approach utilizes the transfer learning technique which reuses information from previous environments to accelerate the search process in the new environment. However, the risk of negative transfer can lead to an erroneous search direction and inefficient use of computational resources. To address this issue, this paper proposes a novel multistage transfer learning DMOEA, named MSTL. The algorithm is designed to enhance the convergence and diversity of the population through the combination of the pre-transfer learning stage and the transfer learning and validation stage when environmental changes occur. In the pre-transfer learning stage, a suitable target domain, source domain, and validation classifier sample sets are generated guided by historical information. In the transfer learning and validation stage, the TrAdaboost technique to transfer knowledge from the target to the source domain, with the results verified by two validation classifiers to mitigate negative transfer. Experimental results demonstrate that our proposed algorithm outperforms four competing algorithms in terms of diversity and convergence on a widely recognized benchmark suite.
Localization algorithm based on the linear least square principle has the advantages of small amount of calculations, and attracts many researchers. In this paper, some algorithms based on linear least square estimati...
详细信息
暂无评论