What does support vector machine (SVM) mean?

Author: The Little Dream, Created: 2017-02-17 10:55:28, Updated: 2017-02-17 11:44:20

What does support vector machine (SVM) mean?

Just recently I saw a post on reddit Please explain Support Vector Machines (SVM) like I am a 5 year old, a word of praise!

  • What is SVM?

    Support vector machine (SVM)

    Of course, first look at the wiki. Support Vector Machines are learning models used for classification: which individuals in a population belong where?

    Well, the story goes like this:

    On Valentine's Day, a long time ago, the hero tries to save his lover, but the devil plays a game with him.

    The devil seems to regularly place balls of two colors on the table, saying: "Hey, did you separate them with a stick?"

    img

    So, what's the big deal about putting it like this?

    img

    Then the devil, again putting more balls on the table, seems to have one of the stands in the wrong camp.

    img

    SVM is about trying to place the stick in the best possible position so that there are as many gaps as possible on both sides of the stick.

    img

    Now, even if the devil throws more balls, the stick is still a good dividing line.

    img

    Then, in the SVM toolkit, there is another more important trick. The Devil sees that the Big Man has learned a trick, so the Devil gives the Big Man a new challenge.

    img

    Now, the hero has no stick to help him separate the two balls, what now? Of course, as in all the martial arts films, the hero hits the table and the ball flies into the air. Then, with the hero's help, the hero grabs a piece of paper and puts it in the middle of the two balls.

    img

    Now, from the devil's point of view, these balls look like they're separated by a curve.

    img

    And then, the boring adults, they call the balls "data", the stick "classifier", the biggest gap trick "optimization", the table called "kernelling", the paper called "hyperplane".

    A video of the demonstration

    See also:

  • It is not a matter of the user's opinion, but the answer is reliable

    img img img img
    img
    img
    img

  • Reply to Han Oliver

    What is SVM?

    SVM - support vector machine, also known as support vector machine, is a supervised learning algorithm that belongs to the category of classification. In data mining applications, it corresponds to and is distinguished from unsupervised clustering. It is widely used in machine learning, computer vision, and data mining. SVM works roughly as shown in Figure 1.

    img

    Let's say we want to divide the real and the empty circles into two categories using the 3-8 line. There are countless lines to accomplish this task. In the SVM, we look for an optimal dividing line so that it has the largest margin on both sides. In this case, the few data points with edge coarsening are called support vectors, which is also the source of the name of this classification algorithm.

    This can be extended to arbitrary n dimensions or even to infinite dimensions, as in Figure 2.

    img

    We got a bunch of data points in a n- dimensional to infinite-dimensional space, Then one can always find a optimal hyperplane which is always in the n-1 dimension.

    And finally, Statistical direction: Support Vector Machines (SVM) wiki:Support vector machine Lesson learned:columbia.eduPage by page I'm also looking forward to a great video demonstration.http://youtu.be/3liCbRZPrZA

Translated from the original German


More