• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Empirical Analysis of Self-Supervised Training Properties

Student: Ildus Sadrtdinov

Supervisor: Ekaterina Lobacheva

Faculty: Faculty of Computer Science

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 10

Year of Graduation: 2021

Self-supervised learning in computer vision tasks has been gaining popularity recently. This paradigm is used to pre-train artificial neural networks on unlabelled data. Modern self-supervised methods considerably outperform ImageNet pre-training. In this work, we compare self-supervised learning, supervised learning, and training on random labels. We show that for each setup, there exist easy and hard examples to be memorized by the neural network. We also demonstrate that training dynamics of self-supervised learning and random labels training are similar.

Full text (added May 17, 2021)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses