<!DOCTYPE html> <!-- saved from url=(0040)http://project.las.ethz.ch/svmdefinition --> <html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> <meta charset="utf-8"> <title>Large Scale Image Classification</title> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <meta name="description" content="Data Mining: Learning from Large Data Sets"> <meta name="author" content="lis2015@inf.ethz.ch"> </head> <body style="zoom: 1;"> <div class="container"> <div class="content"> <h4>Large Scale Image Classification</h4> <p> In this task our goal is to classify images based on their visual content. We provide a labeled dataset containing two images from two classes: Nature and People. While feature selection is a very important part of image classification pipeline, it is out of the scope of this course and won't be part of this task. Instead, we provide a set of features that has been extracted from each image. Your task is to train a model that perfroms well given the feature representation. Furthermore, we will make use of Support Vector Machines and implement the solution using Parallel Stochastic Gradient Descent. </p> <p> The full problem description, includin g dataset description, evaluation metric, grading policy and the data sets is available <a href="http://project.las.ethz.ch/resources/svm.zip"><b>here</b></a>.</p> <p style="font-size:300%">Download and extract data/ from <a href="http://project.las.ethz.ch/resources/svm.zip"><b>here</b></a></p> </div> <footer> <p class="credit"> <small>© Learning and Adaptive Systems Group - Machine Learning Institute - ETH - 2015</small> </p> </footer> </div> <script src="./Large Scale Image Classification_files/wysihtml5.min.js"></script> <script src="./Large Scale Image Classification_files/bootstrap.min.js" type="text/javascript"></script> <script src="./Large Scale Image Classification_files/bootstrap-wysihtml5.min.js"></script> <script src="./Large Scale Image Classification_files/bootstrap-fileupload.min.js" type="text/javascript"></script> </body><div></div><div></div></html>