LIBSVM is an integrated software library for support vector machines, performing support vector classification, (C-SVC, nu-SVC), regression (epsilon-SVR, nu-SVR) and distribution estimation (one-class SVM)

learn more… | top users | synonyms

0
votes
0answers
12 views

Tweaking SVM error rates?

I currently have a one-vs-all SVM setup. Each SVM outputs a score. If I take the maximum score as the correct corresponding class, this gives me FARs of 0.008%. However it also gives me FRRs of about ...
0
votes
0answers
32 views

Handling imbalanced datasets and misclassification costs in SVMs?

I have a dataset with 50 times more negative examples than positive ones. Currently, I am using an oversampling technique to address the imbalance problem. During the model selection stage (i.e. ...
0
votes
1answer
59 views

Decreasing the number of negative examples with respect to the positive examples produces good prediction with SVM

Inroduction We have a binary classification problem and we want to learn a linear SVM (support vector machine). The dataset is composed of: 502 positive examples 5020 negative examples An example ...
0
votes
1answer
15 views

Does LibSVM use Platt Scaling?

I have a binary SVM. I am wondering whether the percentage results that LibSVM gives for each class are Platt scaled?
0
votes
1answer
24 views

Support vector regression (LIBSVM) returns out of range outputs when I use out-of-sample data to predict one step ahead (MATLAB)?

I'm using SVR model in MATLAB R2016a using this option: options_z = ['-q -s 3 -t 2 -c ', C_param, ' -p ', epsilon, ' -g ,Kernel_scale]; I'm optimizing SVR ...
0
votes
0answers
30 views

What should this SVM's input look like?

I'm analyzing curves. Curves consist of a finite number of coordinates x and y. For any 3 two consecutive points of every curve i'm calculating AB angle (the angle between the line AB and the ...
1
vote
2answers
76 views

One-vs-many/One-vs-all - what value to use as probability?

I have constructed SVMs to do a one-vs-many approach to classification. Let's say I have 3 classes and I train 3 SVMs in a one-vs-many format. This gives me 3 SVMs each trained positively on one of a ...
0
votes
0answers
18 views

What is unbalanced data in libsvm?

My full data set (inclusive of both trng & test) has about 350 rows. They are of 3 classes (1,2,3), 1 has about 70, 2 has 125, 3 has 150. The above distribution across the labels is playing a ...
0
votes
1answer
27 views

How to use Cross_validation output of svm-train?

I am getting very poor values with a certain data set I have. I tried to use the -v option of svm-train but later realized that this does not produce any model file for prediction. So what is the ...
0
votes
1answer
24 views

What does the number 'Kernel Option' refer to in SVM?

I read that the performance of some kernel functions in SVM can change if we change the number known as kernel option. For example, this article states that kernel option of value 2 was used, ...
0
votes
1answer
60 views

How to use final svm regression model to predict new values of the dataset

I understand svm_predict function can be used to estimate or predict test output, but the arguments passed are like this svm_estimate = svmpredict(y, X, model); ...
1
vote
1answer
32 views

Use cross-validation trained models for out-of-sample prediction or train a new model using whole data?

I'm using support vector regression(SVR) model and 5-fold cross validation for price prediction. which one is more appropriate after training models for ...
0
votes
0answers
69 views

Any methods to reduce the execution time taken by support vector regression libsvm algorithm?

I am currently working on dataset with 4 input dimensions and 1 output, in each approximately 7000 values hence overall 7000X5 dataset. I chose to use support vector regression and this takes a long ...
0
votes
0answers
17 views

Performance metric for one-class SVM classifier with small number of instances

I am using LibSVM one-class classifier for detecting a particular hand gesture from accelerometer data. I have trained two categories of classifiers: (a) user-dependent: data from the same user is ...
0
votes
0answers
25 views

AUROC for SVM two class classification [duplicate]

I'm trying to compute the ROC and AUROC of a binary svm classification. I already looked up a code in the internet and it's working: ...
1
vote
1answer
50 views

Interpretation of SVM accuracy

I use "LIBSVM" library for classification. If my first class is the same as the second class, for accuracy I will get 50% or 0%. What does it mean to get 30% or 20% of precision?
0
votes
1answer
40 views

Should the target variable be also normalised in SVM?

I understand that normalization is an important preprocessing step for using SVMs, esp. before using non-linear kernels such as the radial basis functions. Should this normalization be applied to the ...
4
votes
1answer
62 views

Machine Learning SVM

If one trains a model using a SVM from kernel data, the resultant trained model contains support vectors. Now consider the case of training a new model using the old data already present plus a small ...
0
votes
0answers
39 views

SKLearn SVC predicting perfectly with random features

I am using an SVC to do binary classification. I am using the rbf kernel and doing leave-one-out cross validation to choose my value of C. I ran the model using my normal features and had a detection ...
2
votes
0answers
4 views

Does the confidence measure of a SVM have a meaning that can be compared across different models?

Suppose I train two SVMs ($m_1$ and $m_2$) on two different (potentially unrelated) problems. I then present datum $d_1$ to model $m_1$ and it outputs a vector $v_1$ of probabilities over the space ...
0
votes
0answers
50 views

How to boost the performance of support vector machine?

I have 4 different data samples: Stage 1: [152 X 27578] Stage 2: [48 X 27578] Stage 3: [48 X 27578] Cancer: [63 X 27578] Each sample are the different stage of cancer in descending order. Here I ...
0
votes
0answers
27 views

Boosting the prediction results in machine learning

I am having the datasets of 152 samples and 151 features.I implemented libSVM algorithm as a classifier. I am getting a classification accuracy just above 55% Is there any way I can boost my ...
1
vote
1answer
45 views

linear kernel SVM

The linear kernel is defined as: $K(x1,x2)=\langle x1,x2\rangle$. I can see that all that this kernel does is to calculate the dot product in the original space of the data. Why is this kernel then ...
0
votes
0answers
67 views

epsilon svr - How to control number of support vector?

I used epsilon SVR in libsvm for predicting time series data, which has approximately 117 rows I used 4 data (t-4, t-3, t-2, t-1) to predict the next data(t). I used gaussian RBF as kernel function. ...
0
votes
0answers
24 views

Is it acceptable to use class probabilities as weights for a weighted average when the bins are numbers 1 to 5?

I have a Multi Class SVM that can predict what class some observation belongs to. There are 5 classes. They are trained for observation that scored 1 to 5. I want the MC-SVM to predict a class for ...
0
votes
0answers
48 views

SVM accuracy after log transformation

I am using SVM (RBF kernel, the LibSVM implementation) to deal with a classification problem. When I use a log 10 transformation for may features values instead of using the default scaling method ...
1
vote
1answer
58 views

Is validation set important in SVM regression analysis

I am new to machine learning and currently reading a paper about a ANN modelling in which they have divided their dataset into Training , validation and testing set. I have carried out some minor SVM ...
0
votes
1answer
26 views

Over sampling imbalance data in SVM

Basically I have a medium size data set (20,000 observations) with only 200 being in group 1, thus an imbalanced data set. My goal is to predict as much group 1 class as possible without sacrificing ...
0
votes
0answers
53 views

High difference between cv rate and classification accuracy in libsvm

edit: nevermind I solved it I am training an svm on a dataset with 5 classes using libsvm. I have a training set and a test set. I am using the easy.py script. The accuracy is a lot worse than the ...
4
votes
1answer
81 views

How probabilities are calculated for SVM model?

I would like to know, how probabilities are calculated in support vector machine. I have used Iris data set and here is my decision values for three "SupportVectorMachine" (please find the PMML below ...
0
votes
0answers
83 views

How to use svm-scale in LIBSVM?

I tried svm-scale -l 0 -u 1 -s range data.data > data_scaled.data but got error: I run the command in windows DOS command box. And I use python interface. Is my ...
0
votes
0answers
51 views

How to extract descriptor values correctly from positive images with HOG? OpenCV

I have extracted descriptor values from positive and negative training set. For positive images, I have cropped objects (bicycle) from positive images, and only extracted descriptor values from them, ...
0
votes
0answers
45 views

Proper scaling of training and test files

I'm trying to classify the web traffic according to the variation of the delay; my goal is to distinguish streaming traffic from downloading traffic. So far, my classes are: ...
1
vote
1answer
24 views

What is the structure of the testing data file in libsvm? [closed]

According to the README file: The format of training and testing data file is: label index1:value1 index2:value2... My training file would be something like this: ...
0
votes
0answers
19 views

libsvm to liblinear migration

I've reached the point when svmlib works too long so I decided to switch to liblinear instead. My question is how to properly transfer parameter training from libsvm to liblinear Say, if I have the ...
-1
votes
1answer
250 views

LibSVM parameter tuning

I am working on LibSVM to classify user comments as negative and positive. I am trying all possible parameter right now however i was not able to find useful information about these parameters Can ...
0
votes
0answers
137 views

How to input data in libsvm for time series prediction using support vector regression?

I am using libsvm to perform time series prediction using support vector regression. Say my series is [11,22,33,44,55,66,77,88,99] and I decide the value of optimum lag to be 2.So from 11 and 22, I ...
0
votes
0answers
42 views

Low cross validation accuracy in L1 Logistic Regression

My question is about the use of L1 Logistic Regression for discriminant analysis (feature selection) in extremely large datasets. I am using liblinear https://www.csie.ntu.edu.tw/~cjlin/liblinear/ for ...
0
votes
0answers
27 views

SVM feature reduction

Suppose task where you need to classify texts using SVM. So before that you need to extract some features from texts. Suppose you pick up many of them and get huge vector. Is there any automatic ...
0
votes
0answers
30 views

libsvm probability seems inverted

I started using libsvm a few days ago. Previously I used OpenCV implementation of SVM, but I needed probabilities on estimation and a read that with libsvm is possible. I started with a simple example ...
0
votes
1answer
93 views

Usage of -b in svm-predict in libsvm

I have to use libsvm for scene-recognition. I have just started reading its usage here in github. I wanted to know what does '-b' exactly does in following usages svm-train -s 0 -b 1 data_file ...
0
votes
0answers
24 views

How to get best possible model in ONE CLASS SVM

I am using one class SVM on a project. I do not have labeled data. I am not sure what should be the judging criteria to identify the best model. Any Suggestions? Thanks in advance!
2
votes
0answers
65 views

Bug in libsvm? Adding a single new example crushes everything! [closed]

I have a set of unbalanced examples, i.e. there is much more negatives than positives. Hence, I decided to use an SVM with weighting. However, when I want to figure out the 'best parameters' (cost and ...
4
votes
2answers
95 views

SVM with non-negative weights

An SVM classifier can be obtained by solving the following, $\arg\min \frac{1}{2}\|W\|_2^2 + C\sum_i \max(0, 1-y_i (W^T\mathbf{x}_i + b))$ where $W$ is the hyperplane (or weights), $b$ is the bias, ...
1
vote
2answers
417 views

How to get probability from the confidence score in SVM

In liblinear library we can get confidence score (the distance between decision hyperplane) in SVM solver for a binary classification problem, but if i want a probability value for membership in any ...
0
votes
0answers
48 views

how can I fixed size of grid search in libsvm?

I'm trying to use libsvm to classify a database that contains 1000 labels using svm one vs rest . My goal is to get out the probabilities for each class and to perform accuracy. I know that the first ...
4
votes
1answer
102 views

How to prove that text is linearly separable?

I sentiment analisys task, for this I used SVM with an rbf kernel and a linear one. The results for the linear kernel were better than the rbf, from this I know that text is linearly separable, but ...
0
votes
0answers
156 views

Bias terms in SVMs

I'm training multi-class classification models using linear SVMs - by learning binary one-vs-all classifiers for each class. To classify a test instance, I evaluate the following equation for each ...
0
votes
1answer
177 views

Only one support vector in a linear svm kernel

I am new to SVM, but I would like to understand certain things. Firstly, when dealing with multiclass classifications, I have a large number of support vectors as proven by R. However, when I run ...
0
votes
0answers
35 views

Lowering C parameter increases the number of support vectors

I know that the C (cost) parameter controls the trade-off between model complexity and misclassification. A large C should increase the error weight, and therefore the model complexity. Nevertheless, ...