Implementing SVM in OS

CHENLei,LUOJun,Yang Xuejun

DOI: https://doi.org/10.3969/j.issn.1000-3428.2001.03.059

2001-01-01

Abstract:This paper researchs which aspect will improve the performance when implementing SVM in OS. First, this paper analyses the kernel of Linux and the key technology of SVM, then gives a project of experiment based on the analyse, at last, does the experiment on a new SVM system and gives some concolusion.

What problem does this paper attempt to address?

Hardware Design of Kernel Functions Operation Suitable for Sensor Systems

JIANG De,QIU Yue,SHEN Hai-bin

DOI: https://doi.org/10.3969/j.issn.1000-9787.2013.09.026

2013-01-01

Abstract:Aiming at trend of technology combining artificial intelligence and sensor,two kinds of hardware implementation methods of support vector machine(SVM) kernel fuction operation are presented.The two methods both support multiple operations of kernel functions and use serial operation style,which avoid the disadvantage of unflexibility to sample amount and excessive sensitivity of area to sample amount in previous structure.The two methods are compared in multiple perspectives,and advantages and disadvantages are discussed.Experimental result shows that one of them suits the situation of critical area limitation and supporting small amount of kernel functions,while the other one suits the situation of high-speed and supporting more amounts of kernel functions.
A Parallel and Scalable Digital Architecture for Training Support Vector Machines

Kui-kang Cao,Hai-bin Shen,Hua-feng Chen

DOI: https://doi.org/10.1631/jzus.c0910500

2010-01-01

Abstract:To facilitate the application of support vector machines （SVMs） in embedded systems,we propose and test a parallel and scalable digital architecture based on the sequential minimal optimization （SMO） algorithm for training SVMs.By taking advantage of the mature and popular SMO algorithm,the numerical instability issues that may exist in traditional numerical algorithms are avoided.The error cache updating task,which dominates the computation time of the algorithm,is mapped into multiple processing units working in parallel.Experiment results show that using the proposed architecture,SVM training problems can be solved effectively with inexpensive fixed-point arithmetic and good scalability can be achieved.This architecture overcomes the drawbacks of the previously proposed SVM hardware that lacks the necessary flexibility for embedded applications,and thus is more suitable for embedded use,where scalability is an important concern.
A Scalable Hardware Implementation of the Support Vector Machine

王嗣平,沈海斌,严晓浪

DOI: https://doi.org/10.3785/j.issn.1008-9497.2009.03.009

2009-01-01

Abstract:This paper presented a new digital structure for support vector machine(SVM) learning,which can get a good performance with less hardware resource while the process speed is not reduced.Also the SVM learning based on the new structure can make itself to suit the number of the sample which is less than the expected.Because of this adaptions and the reduction of the demand for hardware,SVM learning can be applied more widely and conveniently in the embedded system.
SVM KERNEL PARAMETER SELECTION BASED ON IMPROVED GSO

Yang Hai,Ding Yi,Shen Haibin

DOI: https://doi.org/10.3969/j.issn.1000-386x.2015.06.063

2015-01-01

Abstract:Support vector machine (SVM)is a machine learning algorithm with superior performance,the selection of its kernel function parameters greatly affects the modelling accuracy and generalisation ability.This paper proposes an SVMkernel function parameter selection method,it is based on the improved glowworm swarm optimisation (GSO).By improving the glowworm position update formula and introducing brightness feature to the process of moving,the method determines the optimal parameters of SVM kernel function.Experiment shows that the SVM kernel function parameters selected by the method improves classification accuracy while guaranteeing the convergence performance of the classifier,and thus achieves good optimisation effect.
Research on SVM integration for intelligent sensor systen

Gui-long BIAN,Yi DING,Hai-bin SHEN

DOI: https://doi.org/10.13873/J.1000-9787(2014)08-0044-04

2014-01-01

Abstract:Support vector machine( SVM)as the representative of artificial intelligent techniques has been widely used in the intelligent sensor system,however,traditional SVM suffers from the catastrophic forgetting phenomenon,which results in loss of previously learned information,so it is unable to meet the requirements of real-time intelligent sensor system. The strength of Learn++ lies in its ability to learn new data without forgetting previously acquired knowledge,even when the new data introduce new classes. In order to solve the above problem,a Learn++ integration method based on hull vectors is proposed. Experimental results show that the algorithm not only has the ability of incremental learning,improve training speed and reduce the storage size,but also can ensure the classification precision,which meets the current demand of intelligent sensor systems for online learning.
Scalable SVM Processor and Its Application to Nonlinear Channel Equalization

Kuikang Cao,Haibin Shen

DOI: https://doi.org/10.1109/paccs.2009.157

2009-01-01

Abstract:Support vector machines (SVMs) are powerful, state-of-the-art machine learning tools. With the aim to integrate SVM training capability into embedded systems while being able to meet area and performance constraints, a parallel and scalable digital architecture for training SVMs on-line is proposed and implemented on a field-programmable gate array (FPGA). Experiments show that the proposed SVM processor can solve the channel equalization problem effectively with fixed-point arithmetic and exhibits good scalability with respect to the number of cache update units (CUUs). This architecture is particularly suitable to be applied in embedded environments, as designers can easily trade off between area and performance to meet the constraints.
Using Svm To Model And Control Nonlinear Dynamical Systems

Haoran Zhang,Xiaodong Wang,Changjiang Zhang,Jin Wang

2006-01-01

Abstract:This paper firstly provides an short introduction to least square support vector machine (LSSVM), a new class of kernel-based techniques introduced in statistical learning theory and structural risk minimization, then use it to model and control nonlinear systems, in which nonlinear predictive control framework is presented. The predictive control law is derived by a new stochastic search optimization algorithm. Simulation experiments are performed and indicate that the proposed method provides satisfactory performance with excellent generalization property.
SVM Classification:Its Contents and Challenges

Yue Shihong,Li Ping,Hao Peiyi

DOI: https://doi.org/10.1007/s11766-003-0059-5

2003-01-01

Abstract:SVM (support vector machines) have become an increasingly popular tool for machine learning tasks involving classification, regression or novelty detection. In particular, they exhibit good generalization performance on many real issues and the approach is properly motivated theoretically. There are relatively a few free parameters to adjust and the architecture of the learning machine does not need to be found by experimentation. In this paper, survey of the key contents on this subject, focusing on the most well-known models based on kernel substitution, namely SVM, as well as the activated fields at present and the development tendency, is presented.
A VSC Scheme for Linear MIMO Systems Based on SVM

Zhang Yibo,Yang Chunjie,Pi Daoying,Sun Youxian

DOI: https://doi.org/10.1007/11539087_91

2005-01-01

Abstract:A variable structure control (VSC) scheme for linear MIMO systems based on support vector machine (SVM) is developed. By analyzing the characters of linear MIMO system, a VSC scheme based on Exponent Reaching Law is adopted to track desired trajectory. Then one input of the system is trained as the output of SVM, while sliding mode function, differences and other inputs of the system are trained as the inputs of SVM. So one VSC input of the black-box system could be obtained directly by trained SVM after other inputs of the system are selected manually, and recognition of system parameters is avoided. A linear MIMO system is used to prove the scheme, and simulation results show that this scheme has high identification precision and quick training speed.
Study on SVM based model predictive control

Weimin Zhong,Daoying Pi,Youxian Sun

DOI: https://doi.org/10.1109/wcica.2004.1340648

2004-01-01

Abstract:A SVM with linear kernel based single-step predictive model and algorithm are developed after simple discussion of model predictive control. A predictive model is established by non-linear system identification with SVM. And an explicit control law is gotten through the predictive control mechanism. Simulation results show that SVM based predictive control can be applied for non-linear system. And the SVM based single-step predictive of non-linear system has good performance in keeping reference trace and interference-free.
A New Compound Kernel Function for SVM.

Yonghua Mao,Xingshi He,Xiaolin Gui,Ying Guo

DOI: https://doi.org/10.1109/uic-atc-scalcom-cbdcom-iop.2015.236

2015-01-01

Abstract:Support Vector Machines (SVM) is one of most important algorithm in machine learning area. The choice of kernel function can have great influence on classification and approximation ability. Choosing appropriate kernel function and weight parameters is one of the keys to utilize SVM. Single kernel function always has its limitation in the application. We propose a new kernel function based on the analysis about the constitute conditions of the kernel function and the characteristics of different kinds of kernel function-linear compound kernel function, this function not only can reduce the amount of parameters of the kernel function, but also has good learning ability and generalizing ability. And we have tested the effectiveness of the kernel function through simulation.
A geometrical method to improve performance of the support vector machine.

Peter Williams,Sheng Li,Jianfeng Feng,Si Wu

DOI: https://doi.org/10.1109/TNN.2007.891625

2007-01-01

Abstract:The performance of a support vector machine (SVM) largely depends on the kernel function used. This letter investigates a geometrical method to optimize the kernel function. The method is a modification of the one proposed by S. Amari and S. Wu. Its concern is the use of the prior knowledge obtained in a primary step training to conformally rescale the kernel function, so that the separation between the two classes of data is enlarged. The result is that the new algorithm works efficiently and overcomes the susceptibility of the original method.
On-line Support Vector Machine Training Algorithm and Its Application

汪辉,皮道映,孙优贤

DOI: https://doi.org/10.3785/j.issn.1008-973x.2004.12.023

2004-01-01

Abstract:In order to increase training speed of original on-line support vector machine training algorithm, and to deal with the case of margin support vector set being empty, a new algorithm was obtained by strict mathematical deduction based on KKT condition and Lagrangian multiplier method. After a matrix cache was constructed to save data related with kernel function, the method to handle the cache effectively in the algorithm was presented to increase training speed. When the margin support vector set was empty, some sample was added into this set by updating the model bias, so that the training algorithm could run again. Also application of the algorithm to on-line system identification was discussed. The simulation and analysis results show that the modeling precision of the new algorithm is improved for nonlinear and time-varying system,and training speed is increased remarkably compared with that of the original algorithm.
An information-geometrical method for improving the performance of support vector machine classifiers

shunichi amari,si wu

DOI: https://doi.org/10.1049/cp:19991089

1999-01-01

Abstract:The performance of support vector machine (SVM) largely depends on the kernel. There have been no theories concerning how to choose a good kernel in a data-dependent way. As a first step to this important problem, we propose an information-geometrical method of modifying a kernel function to improve the performance of a SVM classifier. The idea is to enlarge the spatial resolution around the separating boundary surface by a conformal mapping. We gave examples of modifying Gaussian Radial Basis Function kernels. Stability of such processes is also known. Simulation results for both artificial and real data turns out to support our idea.
Performance evaluation of kernel functions based on grid search for support vector regression

Xiaoyan Ma,Yanbin Zhang,Yanxia Wang

DOI: https://doi.org/10.1109/ICCIS.2015.7274635

2015-01-01

Abstract:The support vector machine (SVM) has recently been successfully used for classification and regression problems. As the core of SVM, the kernel function is used to map the input vectors into a higher dimensional feature space to realize the non-linear algorithm for original space. The selection of kernel function and corresponding parameters directly affects the learning ability and generalization ability of the support vector regression (SVR). This paper compares the performance of different kernel functions on the energy efficiency data obtained from the UCI repository. The kernel functions used are as follows: Linear, Gaussian, Polynomial, Sigmoid, Inv. Multiquadric and the Semi Local kernel. The influence of parameters on the mean square error of cross validation (MSECV) for each kernel is shown in the paper. The combination of SVR parameters and kernel parameter with minimum MSECV obtained by grid search method based on cross-validation is chosen for SVR model building. The experiment results based on the dataset showed that the Semi Local kernel has the best predictive capability. The mean-squared error of prediction (MSEP) of the Semi Local kernel was 97.13%, 8.26%, 0.96%, 99.95%, 27.36% lower than the value of the Linear, Gaussian, Polynomial, Sigmoid and Inv. Multiquadric kernel.
A new iris recognition approach for embedded system

Hongying Gu,Yueting Zhuang,Yunhe Pan,Bo Chen

DOI: https://doi.org/10.1007/11535409_14

2004-01-01

Abstract:Iris recognition is a prosperous biometric method, but some technical difficulties still exist especially when applied in embedded systems. Support Vector Machine (SVM) has drawn great interests recently as one of the best classifiers in machine learning. In this paper, we develop an iris recognition system using SVM to classify the acquired features series. Even though the SVM outperforms most of other classifiers, it works slowly, which may hinder its application in embedded systems, where resources are usually limited. To make the SVM more applicable in embedded systems, we make several optimizations, including Active Learning, Kernel Selection and Negative Samples Reuse Strategy. Experimental data show that the method presented is amenable: the speed is 5 times faster and the correct recognition rate is almost the same as the basic SVM. This work makes iris recognition more feasible in embedded systems. Also, the optimized SVM can be widely applied in other similar fields.
Study on geometric approach of SVM algorithm-SK algorithm analysis and study

Zhenhua CHANG,Bocheng CHEN,Yingjie LI,Wenhuang LIU,Xuewei YAN

DOI: https://doi.org/10.3778/j.issn.1002-8331.2011.08.044

2011-01-01

Abstract:The geometric approach of the Support Vector Machine(SVM)is a kind of geometric way to find the solution to the problem of the SVM algorithm.Based on its geometric characters,the SK(Schlesinger-Kozinec)algorithm is studied intuitively.It briefly sums up the two convex hulls,based on their relative positions,into five categories,and makes sure their optimizing position got in each computing is mostly at the hull vertices or boundary,it can get to the boundary(the optimization place of the computing)at the first computation.The manual single-step simulation results show that the projection is not always successful for such kind of algorithms in many cases,though it can’t affect the computing result,but can weaken the algorithm efficiency.Based on the analysis,it demonstrates two improving ways for the soft SK algorithm(Backward-SK and Forward-SK methods),and makes some simulation for comparing.The simulation results show that the improved method computing results are almost same as the SK and soft SK ones,but the computing process of improved one is more intuitive.
VLSI Design of Modified Sequential Minimal Optimization Algorithm for Fast SVM Training

Zunchao Li,Yuanfa Wang,Chuang Zheng,Yunhe Guan

DOI: https://doi.org/10.1109/icsict.2016.7998997

2016-01-01

Abstract:In this paper, integrated circuit (IC) implementation of the modified sequential minimal optimization algorithm of supporter vector machine (SVM) is carried out to enable and speed up the learning process of the hardware system. This VLSI circuit system consists of three main operation modules and a control module. A concise and efficient examining module is designed to bring into full play of the advantage of the MSMO algorithm. The presented design is verified on an Altera Cyclone II field-programmable gate array and tested using the publicly available EEG dataset of epilepsy patients for seizure detection. Experiment results show that the designed system is fully functional and speeds up the learning process of SVM.
An Online Support Vector Machine for the Open-Ended Environment.

Xuan Wang,Youlu Xing

DOI: https://doi.org/10.1016/j.eswa.2018.10.027

IF: 8.5

2019-01-01

Expert Systems with Applications

Abstract:Support Vector Machine (SVM) has been widely applied in expert systems. Traditional SVM works as a closed system, which means the parameters of the SVM freeze after training is finished. This strategy leads to the SVM cannot adapt to the online learning environment and also hinders the application of the SVM in expert systems which need to be updated in an online way. Many online SVMs retrain SVM with the current support vectors and new incoming data. However, the retained system usually shows a declined performance due to losing much information about the data. To handle this problem, we propose an Online Support Vector Machine (OSVM). The OSVM includes a Representative Prototype Area (RPA), which retains the representative data of all historical data. In the RPA, each class is maintained by an Online Incremental Feature Map (OIM) which learns a suitable representative set from the stream data automatically, e.g., we do not need to predefine a compression ratio, the OIM will dynamically adapt to the current input data. Retraining is periodically conducted on the updated OIM, which can avoid a declined performance and significantly reduce the computational cost. Variety data sets from different fields are used to test our method and the experimental results demonstrate that the accuracy of OSVM is comparable to that of the state-of-the-art SVM and the speed of OSVM is much faster. Moreover, OSVM is more stable than the other methods in the open-ended learning environment, where stream data from novel classes occur. This implies the OSVM has a good application prospect in online learning expert systems. (C) 2018 Elsevier Ltd. All rights reserved.
Fast Algorithms for Linear and Kernel SVM+.

Wen Li,Dengxin Dai,Mingkui Tan,Dong Xu,Luc Van Gool

DOI: https://doi.org/10.1109/cvpr.2016.248

2016-01-01

Computer Vision and Pattern Recognition

Abstract:The SVM+ approach has shown excellent performance in visual recognition tasks for exploiting privileged information in the training data. In this paper, we propose two efficient algorithms for solving the linear and kernel SVM+, respectively. For linear SVM+, we absorb the bias term into the weight vector, and formulate a new optimization problem with simpler constraints in the dual form. Then, we develop an efficient dual coordinate descent algorithm to solve the new optimization problem. For kernel SVM+, we further apply the l(2)-loss, which leads to a simpler optimization problem in the dual form with only half of dual variables when compared with the dual form of the original SVM+ method. More interestingly, we show that our new dual problem can be efficiently solved by using the SMO algorithm of the one-class SVM problem. Comprehensive experiments on three datasets clearly demonstrate that our proposed algorithms achieve significant speed-up than the state-of-the-art solvers for linear and kernel SVM+.

Implementing SVM in OS

Hardware Design of Kernel Functions Operation Suitable for Sensor Systems

A Parallel and Scalable Digital Architecture for Training Support Vector Machines

A Scalable Hardware Implementation of the Support Vector Machine

SVM KERNEL PARAMETER SELECTION BASED ON IMPROVED GSO

Research on SVM integration for intelligent sensor systen

Scalable SVM Processor and Its Application to Nonlinear Channel Equalization

Using Svm To Model And Control Nonlinear Dynamical Systems

SVM Classification:Its Contents and Challenges

A VSC Scheme for Linear MIMO Systems Based on SVM

Study on SVM based model predictive control

A New Compound Kernel Function for SVM.

A geometrical method to improve performance of the support vector machine.

On-line Support Vector Machine Training Algorithm and Its Application

An information-geometrical method for improving the performance of support vector machine classifiers

Performance evaluation of kernel functions based on grid search for support vector regression

A new iris recognition approach for embedded system

Study on geometric approach of SVM algorithm-SK algorithm analysis and study

VLSI Design of Modified Sequential Minimal Optimization Algorithm for Fast SVM Training

An Online Support Vector Machine for the Open-Ended Environment.

Fast Algorithms for Linear and Kernel SVM+.