Questions tagged [libsvm]
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)
221
questions
1
vote
1
answer
122
views
An OCR model that can be easily improved on the user side
I am building OCR software, for this purpose I trained a model on many types of fonts, the model is SVM but it is not principled.
Now I want the users of the ...
- 71
1
vote
1
answer
72
views
High Cross Validation but low test accuracy on LibSVM
I am solving the problem of detecting swallowing and non-swallowing events from the audio. I labelled the data using Praat software by marking the swallowing and nonswallowing events. I trained the ...
- 23
0
votes
1
answer
229
views
Should I always pick the parameters with the best CV accuracy for GridSearch?
I’m currently training an SVM for multi classification, and to choose the C and gamma parameters, I'm using Grid search combined with k-fold CV.
I get a cross validation accuracy of 99.8%. I am not ...
0
votes
1
answer
258
views
Lasso regression / SVM convergence CPU -> GPU
I have coded a simple supervised ML classification using 10-20K data points for 25 samples. Linear ML models run quickly for example naive Bayes, linear regression and SVM linear on a small multi-core ...
- 113
0
votes
0
answers
130
views
How to understand the _dual_coef_ parameter in sklearn's kernel svm?
I have a small kernel svm code.
...
- 101
1
vote
0
answers
337
views
How can I make sense of alpha values obtained from scikit-learn OC-SVM?
I am building a ML model that uses the OC-SVM for anomaly detection. For our cost function we require the alphas obtained from the OC-SVM. We use the OC-SVM of scikit-learn, which I assume is based on ...
- 11
1
vote
1
answer
45
views
Learning-Agnostic Evaluation of SVM Models
I am at a point where I want to evaluate existing SVM models.
For this task I assume I have:
SVM model (to make it easier let's say it's a scikit-learn one)
Training Dataset that was used to learn (1)...
- 11
1
vote
0
answers
672
views
Code with explanation on using the OpenCV for image classification using the SVM with python from Scratch
Could anyone share links and resources on Image classification using SVM(Support Vector Machine) from Scratch? Also, there should be the use of the OpenCV library.
I got one link, where I am trying to ...
- 667
1
vote
0
answers
431
views
Adjusting the loss function for Support Vector Machines for SVC in sklearn
I have the following problem.
The minimization problem of the SVM that I want to solve is:
$$
\min_{w, b} \frac{1}{2}w^{T}w + \sum^{m}_{i=1}C_{i}xi_{i}
$$
Subject to:
$$
y_{i}(w^{T}x_{i} - b) \geq 1 - ...
- 11
2
votes
0
answers
351
views
How does the python port of libsvm's predict_proba work?
I've followed through the original libsvm code on it's [github][1].
I'm not concerned about the theoretical backfground of how the probability estimates are derived.
All that I care about is how to ...
- 121
0
votes
1
answer
146
views
Why do distances to hyperplane increase with more training samples in a One-Class SVM?
I am using a One-Class SVM from for anomaly detection. I observe that the distance of classified samples increases roughly proportional with the number of training samples. This is true for inliers ...
4
votes
0
answers
448
views
Does distance from the decision boundary suggest higher confidence that the class prediction is correct using SVM?
Does further distance from the decision boundary threshold suggest higher confidence that the class prediction is correct when using SVM with probability estimates enabled?
This is not a question ...
- 295
4
votes
2
answers
792
views
Which algorithm is implemented in sklearn's SVM method?
I'd like to know which exact version of svm is implemented in slearn. The references section on sklearn's svm page cites libsvm package and a paper from 1999 which is about comparing classification ...
3
votes
0
answers
276
views
SVM Scaling problem with One-Class SVM
I'm trying to mess around with a one-class SVM implementation I hacked together from ArduinoSVM.
I'm using an RBF kernel and training the model with just "in" datapoints with sklearn.
First, as is ...
- 31
2
votes
1
answer
183
views
LibSVM - interpreting model output
I am using libSVM on a subset of the MNIST, and I am struggling to interpret the output. I have learned that rho is the bias term, and that sv_coef is the multiplier used to get to the weight term. ...
- 123
1
vote
1
answer
3k
views
What are the weights assigned to the features (coefficients in the primal problem) in SVR? [closed]
In case of linear kernel, how to interpret the weights in the formulations of nu-SVR?
I am using nu-SVR to estimate the parameters in GARCH(1,1) model:$
\sigma_t^2 = \omega + \alpha y_{t-1}^2 + \beta\...
- 13
3
votes
1
answer
154
views
Recover $\rho$ of $\nu$-SVM from e1071 package in R
Given a dataset $\{(x_i,y_i)\}_{i=1}^n$, the primal problem for $\nu$-SVM is:
\begin{align}
&\min_{w,b,\xi,\rho} && \frac{1}{2}w^\top w-\nu\rho+\frac{1}{n}\sum_{i=1}^n\xi_i\\
&\text{...
- 3,140
2
votes
1
answer
75
views
Why setting SVDD's C-parameter to $> 1$ does not affect the result?
Why setting $C>1$ does not affect the result (compared to $C=1$) according to:
https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/#libsvm_for_svdd_and_finding_the_smallest_sphere_containing_all_data
...
- 965
1
vote
0
answers
184
views
Range of Search Space for the hyperparameters of Support Vector Regresssion (SVR)
I need to know "what should be the practical range of c, gamma and epsilon hyper-parameters during grid search optimization in SVR".
The range of dependent variable lies between 1 to 300 with mean ...
- 11
1
vote
1
answer
891
views
How to perform kFold cross validation in Libsvm's precomputed kernel in MATLAB?
I understand that Libsvm provides 'v 10' option for 10-fold cross-validation in...
- 173
0
votes
1
answer
74
views
Formatting 3d data for LIBSVM
I am going to use LIBSVM for classification of data that uses a 'feature window'. Currently my data is in the form of a M x N x 3 matrix - M instances, N features and for each feature I'm considering ...
- 103
5
votes
1
answer
173
views
soft SVM - degenerate case
According to "A Note on Support Vector Machine Degeneracy", Theorem 4, if the dual problem for soft-SVM has a solution with $\alpha_i \in \{0,C\}, \forall i$, then $w=0$ for the primal problem.
In "...
- 1,047
0
votes
1
answer
72
views
SVM decision non linear
As I understand, to perform a decision in a non linear case (using a kernel) I use the following:
$f(x) = sgn(\sum_{i=1}^{n} y_{i} \alpha_{i} \boldsymbol{k}(x,x_{i})+b)$
Where i=1,..n are the ...
- 83
2
votes
2
answers
19k
views
What is the parameter sigma in svm?
What is the influence of the sigma or gamma parameter for the rbf kernel?
if possible a graph for a better understanding
Is epsilon also known as SIGMA?
- 167
4
votes
1
answer
9k
views
What is the parameter nu in oneClass svm?
According to this link, nu specifies the nu parameter for the one-class SVM model. The nu parameter is both a lower bound for the number of samples that are support vectors and an upper bound for the ...
- 167
2
votes
2
answers
372
views
I have 20000 potential features but only 200 instances, how can I do feature selection for an SVM classifier?
I have 20,000 potential features, but only 200 instances, how can I do feature selection for an SVM classifier, especially LIBSVM?
Is there any fast way to do it rather than brute force? Since, ...
- 295
2
votes
0
answers
41
views
For model selection, using n-fold cross-validation, should performance be averaged over the folds or calculated based on all predictions?
In $n$-fold inner-cross-validation, for model selection, should performance be calculated $n$ times and then averaged, or should all predictions be made covering the whole training set, and then ...
- 295
0
votes
0
answers
182
views
Why SVM classifier generates bad results with LDA in Classification of audio data?
I am using SVM classifier to classify my audio data (instrument strokes). I am using LDA as a feature reduction technique to reduce feature dimensions. After applying classifier the results I am ...
- 143
3
votes
1
answer
6k
views
What Are Shrinking Heuristics
I have been working on a project with LibSVM and have noticed there is an option to train the SVM model with "shrinking heuristics" which are used to speed up the classifier training.
After doing ...
- 133
-1
votes
1
answer
57
views
Prepare data for previously trained LibSVM classifier
I trained a LibSVM classifier, with scaled features in the interval $[-1;1]$ with this equation:
$$x'=2\frac{x-\min x}{\max x - \min x}-1$$
I am happy with the trained model and its accuracy on the ...
- 133
0
votes
0
answers
132
views
One-class svm with given true positive rate
Is it always possible to find a one-class SVM with given true positive rate (TPR)?
I used 1-D grid search with cross-validation to find the width parameter of RBF kernel with .90 TPR. The closest ...
- 965
2
votes
2
answers
6k
views
(SVM) Difference between linear kernel and polynomial kernel of degree 1?
I am new to machine learning. Could anyone tell me the difference between linear kernel vs. polynomial kernel of degree 1 wrt SVM (if there is any difference)?
The reason I asked, I am getting ...
1
vote
0
answers
467
views
Augmenting feature vector with bias in SVM: effect on kernel matrix
It is easier to optimise a SVM without an offset term $b$ since we get rid of one of the linear constraints in the dual formulation. If an offset is still needed, it has been suggested that the ...
- 177
6
votes
0
answers
245
views
Optimising dual SVM: why do some authors drop constraints? [duplicate]
In Hastie's Elements of Statistical Learning the dual problem is put as
$$
\begin{align}
\text{arg min}_\alpha \quad &\ \frac{1}{2}\alpha^\top Q\, \alpha_i- \sum_i \alpha_i\\
\text{subject to}\...
- 177
3
votes
1
answer
383
views
Most widely accepted method of SVM model validation?
I am building an SVM based classifier (libsvm) for two distinct balanced classes which are non-linear but separable with an RBF kernel. Both classes have 175 instances each, which must be utilized ...
- 295
0
votes
1
answer
681
views
what's the more common name of "Three-way data splits"?
I'm using Libsvm to train the model but the result is always not so good. now I'm reading this lecture's document:
http://research.cs.tamu.edu/prism/lectures/iss/iss_l13.pdf
the lecture mentioned "...
-1
votes
1
answer
983
views
How to use svm for a large dataset to predict
I am using svm method for prediction of wind power using windspeed. I have a large historical dataset containing timestamp, wind speed and wind power of size 6.74MB. it is a one year data at each 5 ...
0
votes
1
answer
1k
views
How to apply SVM model to new data using libsvm in MATLAB?
Suppose that I have trained and tested an SVM classifier with the following code:
...
- 289
4
votes
2
answers
10k
views
Can someone explain the RBF Kernel to me?
I have read every explanation out there on this but nobody seems capable of explaining this in a way that I am able to understand. For an SVM RBF Kernel we often say that:
But what does x and x' ...
- 143
0
votes
1
answer
74
views
Why is MSE difference between LIBSVM and my codes?
Now I used LIBSVM toolbox to solve a regression problem. However, I found a question that the value of MSE calculated by the toolbox was different with the value of MSE calculated by myself. Did you ...
- 1
1
vote
1
answer
160
views
bias in SVM in a balanced set - effect of sample size?
I have a dataset with many subjects, and for each subject I have 100 samples for an X variable, while Y has exactly 50%/50% cases for label 1/2. I am trying to run cross-validation with SVM and found ...
0
votes
1
answer
168
views
Does LibSVM output a broken SVM in epsilon-svr mode?
I run epsilon-svr on about 1000 samples (RBF kernel), and then calculate the in-sample training errors.
The hyperparameters are found via a search, ranging from 1e-5 to 1e2.
model->eps is a couple ...
- 101
1
vote
1
answer
633
views
Svm-scaling for training and testing using libsvm on matlab [closed]
I want to use SVM-scale for both training and testing on Matlab using libsvm. In libsvm document implement this but using Python and I need for Matlab. how can I do it?
- 11
0
votes
1
answer
1k
views
Why all "nBSV"s are zero in LIBSVM classification outputs?
I'm training a classification SVM for a binary classification model with 20 features (LIBSVM). These are some reports of my model (with different combination of features). Why all ...
- 3,951
0
votes
1
answer
328
views
Do input features need to be in a distribution?
I am currently starting my machine learning project on binary classification and I have been taught that input data features should be in a distribution (although the distribution type itself need not ...
- 21
-1
votes
1
answer
108
views
Odd Behavior in Cutoff of Soft Margin SVM
In soft margin svm, we solve the following quadratic programming problem.
$$
\text{maximize}\ \sum_{i=1}^N \alpha_i-\sum_{i=1}^N\sum_{j=1}^N\alpha_i\alpha_jy_iy_jX_i^tX_j,\\
\text{subject to}\ \sum_{i=...
- 339
1
vote
0
answers
1k
views
Standardizing vs Normalizing data for Support Vector Machines
I am fairly new to the topic of Support Vector Machines, and despite my readings, I still encounter difficulties in understanding the following: why do some people address the need to normalize data ...
- 111
1
vote
0
answers
187
views
Below chance classification after regressing out class-specific signal
I have a signal X that depends on parameters Y (condition) and S (participant). I'm interested in classifying Y based on X, but a) the YxS contingency table of numbers of observations per Y and S ...
- 293
1
vote
0
answers
187
views
Am I correct for cross-validation or validation set in libSVM?
I used libSVM in MATLAB to find whether there is some relationship between the features and label. My steps are:
Divide the dataset in 60% for training, 20% for testing and 20% for validation.
I used ...
- 21
3
votes
2
answers
2k
views
Do SVM suffer from imbalanced class sizes?
Say I have a population with classes c1 and c2 where size(c1) >> size(c2)
Does the SVM suffer from imbalanced classes sizes where I need to manually correct the training procedure?
- 2,015