Skip to content

SophiaVei/Multi-Label---Multi-Instance-Learning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

Multi-Label & Multi-Instance Learning

This work aims to consolidate the objects of learning from data multi-labeling and learning from bags of cases. We are working with the DeliciousMIL dataset, which contains a set of documents, where the each document is considered as a bag of sentences. We are vectorizing them with sklearn's feature_extraction.text module.

Part A
In the first part we will work on the subject of learning from data multiple tags. We apply two learning techniques from multi-label data in the above data, where the representation of each document concerns the words that includes (bag of words). We present and comment on the results of the experiments based on the given train and test splits. Tip: in this type of sparse representations, linear models and/or linear SVMs usually perform well.

Part B
In the second part we will work on the problem of multi instance learning. We focus on the most common of the 20 classes so we can deal with one binary classification problem. We consider that each document is a bag of sentences. We group the sentences of the training set and represent each document based on the groups to which its sentences belong. We try one more model that represents each document based on all its sentences. Wr present and comment on the results of these two approaches.

Releases

No releases published

Packages

No packages published

Languages