---
tags: Paper Survey
---

# Learning Actionable Representations from Visual Observations (Google Brain) (IROS 18)
* [Paper Link](https://arxiv.org/abs/1808.00928)

## Multi-frame Time-Contrastive Network (mfTCN)
* The authors investigate the effectiveness of learning task-agnostic representations for continuous control tasks.


## mfTCN embedings