检索结果-内蒙古大学图书馆

A Comparative Performance of Optimizers and Tuning of Neural Networks for Spoof Detection Framework

INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS 2022年第4期13卷 884-890页

作者： Chadha, Ankita Abdullah, Azween Angeline, Lorita Taylors Univ Sch Comp Sci Subang Jaya 47500 Selangor Malaysia

The breakthroughs in securing speaker verification systems have been challenging and yet are explored by many researchers over the past five years. The compromise in security of these systems is due to naturally sounding synthetic speech and handiness of the recording devices. For developing a spoof detection system, the back-end classifier plays an integral role in differentiating spoofed speech from genuine speech. This work conducts the experimental analysis and comparison of up-to-date optimization techniques for a modified form of Convolutional Neural Network (CNN) architecture which is Light CNN (LCNN). The network is standardized by exploring various optimizers such as Adaptive moment estimation, and other adaptive algorithms, Root Mean Square propagation and Stochastic gradient descent (SGD) algorithms for spoof detection task. Furthermore, the activation functions and learning rates are also tested to investigate the hyperparameter configuration for faster convergence and improving the training accuracy. The counter measure systems are trained and validated on ASV spoof 2019 dataset with Logical (LA) and Physical Access (PA) attack data. The experimental results show optimizers perform better for LA attack in contrast to PA attack. Additionally, the lowest Equal Error Rate (EER) of 9.07 is obtained for softmax activation with SGD with momentum wrt LA attack and 9.951 for SGD with nestrov wrt PA attack.

关键词： Spoof detection speech synthesis voice conversion convolutional neural networks optimizers gradient descent algorithm spoofed speech automatic speaker verification

来源：评论

学校读者我要写书评

暂无评论

An unsupervised neural network to solve transportation problems

引用

Journal of the Institution of Engineers (India), Part PR: Production Engineering Division 2005年第SEPT.期86卷 5-7页

作者： Mouli, K. V. V. Chandra Srinivas, J. Subbaiah, K.V. Department of Mechanical Engineering Gandhi Institute of Technology and Management Visakhapatnam 530 045 India Department of Mechanical Engineering Chaitnaya College of Engineering Visakhapatnam 530 041 India Department of Mechanical Engineering AU College of Engineering Visakhapatnam 530 003 India

This article presents a new two-layer neural network model for predicting the optimum solution to linear programming problems. An energy function that transforms linear programming problem into a non-linear function is developed from the objective and constraints. The learning rule, based on gradient descent algorithm, is employed to get the appropriate weight structure of the neural network. The network is tested with different examples including a transportation problem. The results are compared along with the available solutions.

关键词： Energy function gradient descent algorithm Linear programming problem Penalty function Two-layer neural network

来源：评论

学校读者我要写书评

暂无评论

Modified gradient Linear Discriminant Analysis for Face Recognition

Modified Gradient Linear Discriminant Analysis for Face Reco...

引用

2008 IEEE International Symposium on Knowledge Acquisition and Modeling Workshop(KAM 2008)

作者： Yunhui He Department of Communication Engineering Nanjing University of Information Science and Technology Nanjing,China

In this paper,we propose two kinds of orthogonalization gradient linear discriminant analysis (OGLDA) algorithms to improve the performance of traditional gradient LDA (GLDA) for undersampled problems in face recognition *** the OGLDA1,the orthogonalization procedure is performed on the discriminant vectors to reduce the redundancy among the discriminant features obtained by ***,all obtained discriminative features can equally contribute to classification performance,which significantly improves the performance of GLDA algorithm for face *** the OGLDA2,the discriminant vectors are resolved one by one in each iterative procedure which overcomes the drawbacks of high computational cost and numerical instability existing in the OGLDA1 *** the orthogonalization procedure is applied in the proposed OGLDA methods,the computational stability is improved *** effectiveness of the proposed methods is verified in the experiments on the standard face image databases.

关键词： face recognition linear discriminant analysis orthogonalization gradient descent algorithm

来源：评论

学校读者我要写书评

暂无评论

A PSO-GD-BASED HYBRID algorithm FOR GENERAL FUZZY MEASURE DETERMINATION

A PSO-GD-BASED HYBRID ALGORITHM FOR GENERAL FUZZY MEASURE DE...

引用

2009 International Conference on Machine Learning and Cybernetics(2009机器学习与控制论国际会议)

作者： HUAN-YU ZHAO XI-ZHAO WANG Key Lab.for Machine Learning and Computational Intelligence College of Mathematics and Computer Science Hebei University Baoding 071002 Hebei China

Determining fuzzy measure from data is an important topic in some practical applications. Some computing techniques are adopted, such as particle swarm optimization (PSO) and gradient descent algorithm (GD), to identify fuzzy measure. However, there exist some limitations. In this paper, we design a hybrid algorithm called CDPSO, through introducing GD to PSO for the first time. This algorithm has the advantages of GD and PSO, and avoids the disadvantages of them. Theoretical analysis and experimental results verify this, and show that GDPSO is effective and efficient.

关键词： Fuzzy measure Fuzzy integral Particle swarm optimization gradient descent algorithm

来源：评论

学校读者我要写书评

暂无评论

A Method of Dynamic System Identification based on Memory RBF Network

A Method of Dynamic System Identification based on Memory RB...

引用

第28届中国控制与决策会议

作者： Qiang Lv Yang Zhang Huican Lin Academy of Armored Force Engineering

ISBN: (纸本)9781467397155

A recursive network is proposed by introducing memory neurons based on the RBF *** to the current output value of the network is related to the past input value in the network,the network will be able to identify the dynamics of the system without the need of explicitly feedback of input and output in the ***,the network is able to identify a system has an unknown order or an unknown *** algorithm and the theoretical rationality are provided in this *** validity of the method is verified by simulation of dynamic system identification,and it is of great significance in the field of adaptive control.

关键词： RBF Network System Identification Recursive Network gradient descent algorithm

来源：评论

学校读者我要写书评

暂无评论

Adaptive On-Line Optimizing the Gaussian Kernel for Classification based on the Kernel Target Alignment

Adaptive On-Line Optimizing the Gaussian Kernel for Classifi...

引用

2015 IEEE International Conference on Computer and Communications(ICCC 2015)

作者： Yulin Xiao Shangping Zhong College of Mathematics and Computer Science Fuzhou University

Kernel target alignment is a very efficient evaluation criterion. It has been widely applied in kernel optimization. However the traditional kernel methods that based on the Kernel target alignment optimize the kernel function mainly with batch gradient descent algorithm. This kind of methods has to scan through the entire training set at each step, which is much too costly. The On-line learning algorithm exactly solve above problem. At each step, on-line learning algorithm only need one example then discarded after learning, which make on-line learning algorithm fast, simple, and often make few statistical assumptions. Thus, in this paper, we propose a novel method to optimize the Gaussian kernel with on-line learning. We formulate the objective criterion for kernel optimization based on kernel target alignment. The objective criterion can be proved to have a determined global minimum point. Then, we use the on-line learning algorithm to optimize the formulated kernel function. In addition, in order to get an appropriate learning rate for the algorithm to accelerate the convergence rate, we use an adaptive rate learning method to optimize the kernel function. Finally, we evaluate the empirical performance of the proposed kernel optimization method on ten diverse datasets. The experimental results show that the proposed method is more effective than the state-of-the-art kernel optimization algorithms.

关键词： kernel target alignment adaptive on-line learning kernel optimization kernel methods gradient descent algorithm

来源：评论

学校读者我要写书评

暂无评论

IMPROVED ONLINE IDENTIFICATION OF ACOUSTIC MISO SYSTEMS BASED ON SEPARATED INPUT SIGNAL COMPONENTS

IMPROVED ONLINE IDENTIFICATION OF ACOUSTIC MISO SYSTEMS BASE...

引用

IEEE International Conference on Acoustics, Speech and Signal Processing

作者： Philipp Thune Gerald Enzner Institute of Communication Acoustics (IKA) Ruhr-Universit?t Bochum Germany

ISBN: (纸本)9781479903573

Creating an immersive listening experience and providing the audience with improved spatial realism is the goal of many adaptive audio reproduction techniques such as room equalization or crosstalk cancellation. The majority of these approaches currently relies on acoustic impulse responses (AIRs) that have been measured prior to the actual audio reproduction. In order to maintain a high degree of adaptivity, however, the AIRs need to be estimated online during the reproduction process, which turns out to be a severely ill-conditioned problem due to the high inter-channel correlation of the loudspeaker signals. In this paper, we present a novel approach to MISO system identification with realistically correlated excitation. Based on the idea of separate treatment of correlated and uncorrelated signal components, we propose two extended filter structures for gradient-descent-based adaptive system identification and provide theoretical analysis and experimental validation of their effectiveness.

关键词： room acoustics multichannel acoustic system identification nonuniqueness gradient descent algorithm room acoustics multiple-input single-output signals ENTERING steepest descent algorithm System identification Audio

来源：评论

学校读者我要写书评

暂无评论

Fast algorithm of Evolutional Learning Neural Network

Fast Algorithm of Evolutional Learning Neural Network

引用

2012 International Conference on Intelligent System Design and Engineering Applications(2012年智能系统设计与工程应用国际会议 ISDEA 2012)

作者： Zhong-hua Xu Wei-bing Chen Wei-feng Yang Fengnian Liu School of Computer and ommunication Hunan University of Technology Zhuzhou Hunan 412008 China

ISBN: (纸本)9781457721205

The neural networks have been widely applied to optimum calculation and solution of complicated problems. In particular, the evolutional learning neural network has better characteristic and higher precision than other neural networks. But the slow computational rate of evolutional learning and the local-optimum of the evolutional learning neural network seriously influence its application. In this paper, the fast algorithm combining the gradient descent algorithm with the evolutional learning algorithm can effectively solve above problems. This neural network has been extensively applied.

关键词： Evolutional learning Neural networks gradient descent algorithm

来源：评论

学校读者我要写书评

暂无评论

Micro-IMU-Based Motion Tracking System for Virtual Training

Micro-IMU-Based Motion Tracking System for Virtual Training

引用

第三十四届中国控制会议

作者： Yang Zhang Yunfeng Fei Lin Xu Guangyi Sun Institute of Robotics and Automatic Information System Nankai University Tianjin Key Laboratory of Intelligent Robotics Engineering Design and Research Institute of the Second Artillery Corps.

ISBN: (纸本)9781467374439

This paper presents the development of a low cost wireless real-time inertial body tracking system for virtual training. The system is designed to provide highly accurate human body motion capture and interactive three-dimensional(3-D) avatar steering, by combining low cost MEMS inertial measurement units(IMUs), wireless body sensor network(BSN), and Unity 3D virtual reality game engine. First, several wearable MEMS IMU sensors are placed on user's body and limbs according to human skeletal action, and each sensor performs a 9 degrees of freedom(DOF) tracking at a high-speed update rate. Second, a Zigbee-based BSN is designed to support up to 20 MEMS IMU sensor nodes data transmission at 50 Hz sampling frequency. All collected sensors' data are loaded to a Matlab-based PC program by means of serial port. In order to accurately estimate the local orientation of each IMU sensor, an optimized gradient descent algorithm is implemented. The algorithm uses a quaternion representation, which allows accelerometer and magnetometer data to be fused to compute the gyroscope measurement error as a quaternion derivative. Finally, the estimated orientation data by fusion algorithm are imported to a virtual environment, consisting of the 3-D virtual skeletal representation and the virtual scene for specific training. Experimental results indicate that the system achieves < 1o static RMS error and <2o dynamic RMS error. The systems further expand the usability of low cost body tracking solution to virtual training in virtual environments.

关键词： micro inertial measurement unit(IMU) motion capture virtual training gradient descent algorithm wireless sensor network

来源：评论

学校读者我要写书评

暂无评论

Effective Control for Wireless Sensor and Mobile Actuator Network in Regulation of Environmental Density Function

Effective Control for Wireless Sensor and Mobile Actuator Ne...

引用

IEEE/ASME International Conference on Advanced Intelligent Mechatronics

作者： Mu-Tai Lin Yen-Chen Liu Department of Mechanical Engineering National Cheng Kung University Tainan Taiwan

This paper proposes an environmental regulating system by using a multi-robot system in cooperation with wireless sensor networks. Wireless sensors pre-deployed in the environment and embedded in mobile robots can gather sensory information to estimate the distribution of environmental variable. The environment distribution is illustrated by a density function which is constructed by Gaussian mixture model based on Expectation Maximization (EM) algorithm to estimate real condition. Subsequently, a gradient decent coverage control is proposed to drive the multi-robot system to cover the optimal distribution based on the estimated density function. Meanwhile, the actuators embedded on mobile robots are designed to regulate the environment to achieve a desired value of density function. Numerical examples are illustrated to show the performance of the proposed control system.

关键词： Wireless sensor and actuator networks (WSANs) Coverage control Expectation maximization estimation gradient descent algorithm Lyapunov-based controller

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：