Time-Contrastive Networks : Self-Supervised Learning from Pixels

By Pierre Sermanet, Corey Lynch, Yevgen Chebotar, Jasmine Hsu, Eric Jang, Stefan Schaal, Sergey Levine. 2017

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Representation & Self-Supervised Learning

1. Sergey Levine
2. Pieter Abbeel
3. Phillip Isola
4. Yonglong Tian
5. Pierre Sermanet
6. Dilip Krishnan
7. Corey Lynch
8. Eric Jang

Citations in the corpus

• Contrastive Multiview Coding (2019, background): “Time contrastive networks [53] use a triplet loss framework to learn representations from aligned video sequences of the same scene, taken by different video cameras.”
• Time-Contrastive Networks: Self-Supervised Learning from Video (2017, methodology): “In Table I we select each model based on the its lowest validation loss, while in Table IV of [40] we select based on a classification score from a small validation set labeled with the 5 attributes described earlier.”
• Evolving Losses for Unlabeled Video Representation Learning (2019, background): “Related Works There are many tasks for self-supervised learning such as predicting if frames appear in order [6, 5, 7], reconstruction or prediction of future frames [9], timecontrastive learning [3, 8].”