Graph cut criterion has been proven to be robust and applicable in clustering problems. In this paper the graph cut criterion is applied to construct a supervised dimensionality reduction. A new graph cut, scaling cut...
详细信息
In this paper,a novel algorithm for color image enhancement is *** proposed method,which is based on Retinex theory and total variational framework,improves the original variational method by using a TV penalty term t...
详细信息
In this paper,a novel algorithm for color image enhancement is *** proposed method,which is based on Retinex theory and total variational framework,improves the original variational method by using a TV penalty term to force spatial smoothness on the reflectance *** split Bregman algorithm is employed to solve the proposed *** practical experiments,it is verified that our method obtains better enhancement performance and much better calculation efficiency.
Recently, Gutiérrez-Naranjo and Leporati considered performing basic arithmetic operations on a new class of bioinspired computing devices - spiking neural P systems (for short, SN P systems). However, the binary...
详细信息
This paper presents a robust tracking algorithm for infrared objects in the image sequence, which is based on particle filer. Particle filter is a powerful tool for tracking especially in non-Gaussian condition, but t...
详细信息
Diverse modeling frameworks have been utilized with the ultimate goal of translating brain cortical signals into prediction of visible behavior. The inputs to these models are usually multidimensional neural recording...
详细信息
Traditional recurrent neural networks are composed of capacitors, inductors, resistors, and operational *** neural networks are constructed by replacing resistors with memristors. This paper focuses on the memory anal...
详细信息
Traditional recurrent neural networks are composed of capacitors, inductors, resistors, and operational *** neural networks are constructed by replacing resistors with memristors. This paper focuses on the memory analysis,i.e. the initial value computation, of memristors. Firstly, we present the memory analysis for a single memristor based on memristors’ mathematical models with linear and nonlinear ***, we present the memory analysis for two memristors in series and parallel. Thirdly, we point out the difference between traditional neural networks and those that are memristive. Based on the current and voltage relationship of memristors, we use mathematical analysis and SPICE simulations to demonstrate the validity of our methods.
The sufficient and necessary conditions for Lyapunov stability of the zero equilibrium point of Lorenz system are discussed, and some brief criteria are presented for globally exponential stability, globally asymptoti...
详细信息
The sufficient and necessary conditions for Lyapunov stability of the zero equilibrium point of Lorenz system are discussed, and some brief criteria are presented for globally exponential stability, globally asymptotical stability and instability. Furthermore, the behavior of the non-zero equilibrium point of Lorenz system is also investigated, and several sufficient and necessary conditions are provided for locally exponential stability and instability. The established theorems in this paper develop and extend the existing achievements on Lyapunov stability of Lorenz system. In conclusion, by applying the obtained results, some less conservative feedback-control laws are derived to guarantee the globally exponential stability of the chaos control of Chen system, Lu¨ system, Yang-Chen system and Yu-Xia Li system.
SiGe materials have become a research hotspot due to their important applications in semiconductor devices, especially in optoelectronic and high-speed electronic devices. In this study, based on molecular dynamics si...
SiGe materials have become a research hotspot due to their important applications in semiconductor devices, especially in optoelectronic and high-speed electronic devices. In this study, based on molecular dynamics simulations, the influence of the Si buffer layer on the quality of films in Ge/Si/SiGe heterostructures is investigated. By simulating the growth process of the Ge/Si/SiGe heterostructure, a deposition model based on Ge(100) substrates is established. Inspired by the concept of reverse gradient buffer layers, Si buffer layers are directly grown on Ge substrates, followed by the deposition of SiGe films. This study primarily investigates the effects of the growth temperature and deposition thickness of the Si buffer layer on the quality of SiGe films. Based on the deposition parameters identified as suitable under the current simulation conditions (620 °C, 9.7 nm), the influence of the buffer layer on SiGe films with varying Ge compositions is further analyzed. The results show that the dislocations and stacking faults formed in the Si buffer layer effectively relieve the stress caused by lattice mismatch, thus improving the crystal quality of the subsequent SiGe films. This study provides theoretical insights into the Ge/Si/SiGe heterostructure film growth process, which helps enhance the quality of SiGe films and expands their applications in semiconductor devices.
In this paper, we propose a novel regularization algorithm that is introduced as a penalty term to the loss function. Differing from conventional L1 and L2 regularization methods, our approach does not aim to diminish...
In this paper, we propose a novel regularization algorithm that is introduced as a penalty term to the loss function. Differing from conventional L1 and L2 regularization methods, our approach does not aim to diminish the weights of individual neurons or enforce sparsity by driving certain neurons to zero. Instead, it functions by increasing the differences between neurons and enhancing the diversity of neurons within each layer. Our method incorporates ensemble learning techniques by treating the layer weight matrix as a collective learning model, where each neuron serving as a weak learner within the layer. The proposed algorithm improves the performance of DCNN by simultaneously considering the distance between multiple filters in the same layer. This algorithm reduces the redundancy of the parameter layer filters in DCNN and enhances its robustness. The penalty term proposed by our algorithm dynamically adjusts its value in a cyclical manner, compelling the neural network to navigate away from its current gradient state. In the parameter space, different weights correspond to different locations. The proposed algorithm quantifies the distance between neurons and iteratively increases the distance between neurons during thereby encouraging greater diversity within the network. Experimental evaluations demonstrate the effectiveness of our algorithm in enhancing neural network performance without requiring adjustments to other hyper-parameters.
The immune system’s ability to adapt its B cells to new types of antigen is powered by processes known as clonal selection and affinity maturation. When the body is exposed to the same antigen,immune system usually c...
详细信息
The immune system’s ability to adapt its B cells to new types of antigen is powered by processes known as clonal selection and affinity maturation. When the body is exposed to the same antigen,immune system usually calls for a more rapid and larger response to the antigen,where B cells have the function of negative adjustment. Based on the clonal selection theory and the dynamic process of immune response,two novel artificial immune system algorithms,secondary response clonal programming algorithm (SRCPA) and secondary response clonal multi-objective algorithm (SRCMOA),are presented for solving single and multi-objective optimization problems,respectively. Clonal selection operator (CSO) and secondary response operator (SRO) are the main operators of SRCPA and SRCMOA. Inspired by the clonal selection theory,CSO reproduces individuals and selects their improved maturated progenies after the affinity mat-uration process. SRO copies certain antibodies to a secondary pool,whose members do not participate in CSO,but these antibodies could be activated by some external stimulations. The update of the secondary pool pays more attention to maintain the population diversity. On the one hand,decimal-string representation makes SRCPA more suitable for solving high-dimensional function optimiza-tion problems. Special mutation and recombination methods are adopted in SRCPA to simulate the somatic mutation and receptor edit-ing process. Compared with some existing evolutionary algorithms,such as OGA/Q,IEA,IMCPA,BGA and AEA,SRCPA is shown to be able to solve complex optimization problems,such as high-dimensional function optimizations,with better performance. On the other hand,SRCMOA combines the Pareto-strength based fitness assignment strategy,CSO and SRO to solve multi-objective optimization problems. The performance comparison between SRCMOA,NSGA-Ⅱ,SPEA,and PAES based on eight well-known test problems shows that SRCMOA has better performance in converging to approximate Pareto
暂无评论