Let the machine teach people to learn more effectively: becoming the expert-interactive multi-class machines teaching

This is a very interesting article from CVPR2015, which studies how machines can automatically teach students to classify their learning goals. This paper develops a machine learning method, which optimizes the method and skill of machine instruction and improves the efficiency of learning target classification through the process of human-computer interaction. As shown in the following:

The process by which the machine teaches people to learn

1, the machine first show a picture A to people, do not tell the picture of the real category of objects;

2, the person to judge the picture, give a category answer ry;

3, the machine based on the results of human judgment to update the algorithm model m to optimize;

4, then the machine will picture a real category Ty to tell people;

5, repeat the above steps until the end of the study.

The result of the experiment is that the proposed method is effective, as shown in:

Summarize

The forehead enlightened a lot, this method in the online education, human-computer interaction has great application value. Thesis contribution:

1, automatic learning, no human intervention. Similar to the back of words, in the back of the machine has been unconsciously learned how to better teach people/students.

2, because of the effectiveness of teaching, learning time will be shortened, that is, efficiency will improve.

3. The author provides a website that allows you to learn more about the target category and apply it to other visual tasks heavy.

Literature

: http://arxiv.org/abs/1504.07575

Paper Project Address: http://visual.cs.ucl.ac.uk/pubs/interactiveMachineTeaching/

Let the machine teach people to learn more effectively: becoming the expert-interactive multi-class machines teaching