Page 10 of 13 for All Posts - Page 10

Leveraging Compositionality for One-Shot Imitation Learning

We introduce a class of algorithms that solve long-horizon one-shot imitation learning by leveraging compositional priors.

SAIL at ICLR 2020: Accepted Papers and Videos

Compiled by Sharon Li, Karan Goel and Andrey Kurenkov

All the great work from the Stanford AI Lab accepted at ICLR 2020, in one place.

Automating Data Augmentation: Practice, Theory and New Direction

Sharon Y. Li

Data augmentation is a de facto technique used in nearly every state-of-the-art machine learning model. In this blog post, we provide an overview of recent work on the practice, theory and new direction of data augmentation research.

Sequential Problem Solving by Hierarchical Planning in Latent Spaces

Kuan Fang

We propose a hierarchical planning algorithm in learned latent spaces. Our method uses deep generative models to prioritize promising actions for sampling-based planning.

BERT, ELMo, & GPT-2: How Contextual are Contextualized Word Representations?

Kawin Ethayarajh

Replacing static vectors with contextualized word representations has led to significant improvements on virtually every NLP task. In this blog post we study the geometric properties of contextualized word representations and find surprising conclusions.

When Humans Aren’t Optimal: Robots that Collaborate with Risk-Aware Humans

Minae Kwon and Dylan Losey

To create human-like robots, we need to understand how humans behave. We present a modeling approach enables robots to anticipate that humans will make suboptimal choices when risk and uncertainty are involved.

Text Feature Selection for Causal Inference

Reid Pryzant

Does adding a theorem to any paper increase its chance of acceptance? Which writing style in an email would get the fastest response, regardless of its subject? We introduce algorithms and a python package for measuring the importance of ngram features on an outcome, while controlling for confounds.

RoboNet: A Dataset for Large-Scale Multi-Robot Learning

Sudeep Dasari

We introduce RoboNet, an open database for sharing robotic experience, and study how this data can be used to learn generalizable models for vision-based robotic manipulation. We find that pre-training on RoboNet enables faster learning in new environments compared to learning from scratch.

Controlling Assistive Robots with Learned Latent Actions

Dylan Losey

We want to make it easier for humans to teleoperate dexterous robots. We present a learning approach that embeds high-dimensional robot actions into an intuitive, human-controllable, and low-dimensional latent space.

RoboTurk: Human Reasoning and Dexterity for Large-Scale Dataset Creation

Ajay Mandlekar

We built a system that enables collecting large-scale robot manipulation datasets with human supervision and used it to collect the largest robot dataset ever collected via teleoperation.