e1071: CVE returns linear separation, Hold-out returns large error
I’m using the Adult dataset that can be found here: http://archive.ics.uci.edu/ml/datasets/Adult After taking a sample of the dataset, I use the svm function of e1071 to obtain the accuracy with a...
View ArticleWhat is the difference between Machine Learning and Deep Learning?
OK, I know there is a lot of topic regarding this in the internet, and trust me, I’ve googled it. But things are getting more and more confused for me. From my understanding, Deep Learning (DL) is kind...
View ArticleAccuracy on the test set do not change. Why?
I train a SVM classifier using 36 features. If I use all the features, the train accuracy is about 0.96, the test accuracy is about 0.77. Then I change the number of features. The train accuracy drops...
View ArticleAre there kernel-based one-class sparse kernel-based outlier detection...
I have a commercial outlier detection problem in moderate dimension (8-25). We have a limited number of true positive tags and can roughly evaluate performance of various methods. So far, the 1-class...
View ArticleNon-probabilistic vs probabilistic frameworks for decision theory in metric...
I have a task to make a decision, say to classify an object as $X$ or $ overline X$. However, $overline X$ usually means everything else and you only have positive examples of $X$ and not so many...
View ArticleOverfitting in One-Class SVM
In a one-class SVM model, would a low value of $nu$ be considered over-fitting the model or would a large value of $nu$ be considered over-fitting? I’m very confused as, in the latter case, as $nu...
View ArticleCan we compare classifier scores in one-vs-all/one-vs-many?
In a system where we perform multi-class classification via a one-vs-all technique, are two scores comparable? E.g.: If I have 0.5 and 0.6 on two different classifiers, is it possible to say that the...
View ArticleTweaking 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...
View Article