Stanford AI Lab Papers and Talks at ICLR 2021

The International Conference on Learning Representations (ICLR) 2021 is being hosted virtually from May 3rd – May 7th. We’re excited to share all the work from SAIL that’s being presented, and you’ll find links to papers, videos and blogs below. Feel free to reach out to the contact authors directly to learn more about the work that’s happening at Stanford!

List of Accepted Papers

Adaptive Procedural Task Generation for Hard-Exploration Problems


Authors: Kuan Fang, Yuke Zhu, Silvio Savarese, Li Fei-Fei

Contact: kuanfang@stanford.edu

Links: Paper | Video | Website

Keywords: curriculum learning, reinforcement learning, procedural generation


Anytime Sampling for Autoregressive Models via Ordered Autoencoding


Authors: Yilun Xu, Yang Song, Sahaj Garg, Linyuan Gong, Rui Shu, Aditya Grover, and Stefano Ermon

Contact: ylxu@mit.edu

Links: Paper

Keywords: autoregressive models, anytime algorithm, sampling


Concept Learners for Few-Shot Learning


Authors: Kaidi Cao*, Maria Brbić*, Jure Leskovec

Contact: kaidicao@cs.stanford.edu, mbrbic@cs.stanford.edu

Links: Paper | Website

Keywords: few-shot learning, meta learning


Conditional Negative Sampling for Contrastive Learning of Visual Representations


Authors: Mike Wu, Milan Mosse, Chengxu Zhuang, Daniel Yamins, Noah Goodman

Contact: wumike@stanford.edu

Links: Paper

Keywords: contrastive learning, negative samples, mutual information


Cut out the annotator, keep the cutout: better segmentation with weak supervision


Authors: Sarah Hooper, Michael Wornow, Ying Hang Seah, Peter Kellman, Hui Xue, Frederic Sala, Curtis Langlotz, Christopher Ré

Contact: smhooper@stanford.edu

Links: Paper

Keywords: medical imaging, segmentation, weak supervision


Evaluating the Disentanglement of Deep Generative Models through Manifold Topology


Authors: Sharon Zhou, Eric Zelikman, Fred Lu, Andrew Y. Ng, Gunnar E. Carlsson, Stefano Ermon

Contact: sharonz@cs.stanford.edu

Links: Paper | Website

Keywords: generative models, evaluation, disentanglement


Heteroskedastic and Imbalanced Deep Learning with Adaptive Regularization


Authors: Kaidi Cao, Yining Chen, Junwei Lu, Nikos Arechiga, Adrien Gaidon, Tengyu Ma

Contact: kaidicao@cs.stanford.edu

Links: Paper

Keywords: deep learning, noise robust learning, imbalanced learning


How Does Mixup Help With Robustness and Generalization?


Authors: Linjun Zhang, Zhun Deng, Kenji Kawaguchi, Amirata Ghorbani, James Zou

Contact: jamesz@stanford.edu

Links: Paper

Keywords: mixup, adversarial robustness, generalization


In-N-Out: Pre-Training and Self-Training using Auxiliary Information for Out-of-Distribution Robustness


Authors: Sang Michael Xie*, Ananya Kumar*, Robbie Jones*, Fereshte Khani, Tengyu Ma, Percy Liang

Contact: xie@cs.stanford.edu

Links: Paper | Website

Keywords: pre-training, self-training theory, robustness, out-of-distribution, unlabeled data, auxiliary information, multi-task learning theory, distribution shift


Interactive Weak Supervision: Learning Useful Heuristics for Data Labeling


Authors: Benedikt Boecking, Willie Neiswanger, Eric Xing, Artur Dubrawski

Contact: neiswanger@cs.stanford.edu

Links: Paper | Website

Keywords: weak supervision, active learning, interactive learning, data programming, level set estimation


Language-Agnostic Representation Learning of Source Code from Structure and Context


Authors: Daniel Zügner, Tobias Kirschstein, Michele Catasta, Jure Leskovec, Stephan Günnemann

Contact: pirroh@cs.stanford.edu

Links: Paper | Blog Post | Website

Keywords: transformer; source code; ml4code


Learning Energy-Based Models by Diffusion Recovery Likelihood


Authors: Ruiqi Gao, Yang Song, Ben Poole, Ying Nian Wu, and Diederik P. Kingma

Contact: ruiqigao@ucla.edu

Links: Paper

Keywords: energy-based models, diffusion score models, generative modeling


Learning from Protein Structure with Geometric Vector Perceptrons


Authors: Bowen Jing, Stephan Eismann, Patricia Suriana, Raphael John Lamarre Townshend, Ron Dror

Contact: bjing@cs.stanford.edu, seismann@cs.stanford.edu

Links: Paper | Website

Keywords: structural biology, graph neural networks, proteins, geometric deep learning


MONGOOSE: A Learnable LSH Framework for Efficient Neural Network Training


Authors: Beidi Chen, Zichang Liu, Binghui Peng, Zhaozhuo Xu, Jonathan Lingjie Li, Tri Dao, Zhao Song , Anshumali Shrivastava , Christopher Ré

Contact: beidic@stanford.edu

Award nominations: Oral

Links: Paper | Video | Website

Keywords: efficient training; locality sensitive hashing; nearest-neighbor search;


Model Patching: Closing the Subgroup Performance Gap with Data Augmentation


Authors: Karan Goel*, Albert Gu*, Sharon Li, Christopher Re

Contact: kgoel@cs.stanford.edu

Links: Paper | Blog Post | Video | Website

Keywords: data augmentation, robustness, consistency training


Nearest Neighbor Machine Translation


Authors: Urvashi Khandelwal, Angela Fan, Dan Jurafsky, Luke Zettlemoyer, Mike Lewis

Contact: urvashik@stanford.edu

Links: Paper

Keywords: nearest neighbors, machine translation


On the Critical Role of Conventions in Adaptive Human-AI Collaboration


Authors: Andy Shih, Arjun Sawhney, Jovana Kondic, Stefano Ermon, Dorsa Sadigh

Contact: andyshih@cs.stanford.edu

Links: Paper | Blog Post | Website

Keywords: multi-agent systems, human-robot interaction


PMI-Masking: Principled masking of correlated spans


Authors: Yoav Levine, Barak Lenz, Opher Lieber, Omri Abend, Kevin Leyton-Brown, Moshe Tennenholtz, Yoav Shoham

Contact: shoham@cs.stanford.edu

Award nominations: Spotlight selection

Links: Paper

Keywords: masked language models, pointwise mutual information (pmi)


Score-Based Generative Modeling through Stochastic Differential Equations


Authors: Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon, Ben Poole

Contact: yangsong@cs.stanford.edu

Award nominations: Outstanding Paper Award

Links: Paper | Blog Post | Website

Keywords: generative modeling, stochastic differential equations, score matching, inverse problems, likelihood


Selective Classification Can Magnify Disparities Across Groups


Authors: Erik Jones*, Shiori Sagawa*, Pang Wei Koh*, Ananya Kumar, Percy Liang

Contact: erjones@cs.stanford.edu

Links: Paper

Keywords: selective classification, group disparities, log-concavity, robustness


Theoretical Analysis of Self-Training with Deep Networks on Unlabeled Data


Authors: Colin Wei, Kendrick Shen, Yining Chen, Tengyu Ma

Contact: colinwei@stanford.edu

Links: Paper

Keywords: deep learning theory, domain adaptation theory, unsupervised learning theory, semi-supervised learning theory


Viewmaker Networks: Learning Views for Unsupervised Representation Learning


Authors: Alex Tamkin, Mike Wu, Noah Goodman

Contact: atamkin@stanford.edu

Links: Paper | Blog Post

Keywords: contrastive learning, domain-agnostic, pretraining, self-supervised, representation learning


Practical Deepfake Detection: Vulnerabilities in Global Contexts


Authors: Yang Andrew Chuming, Daniel Jeffrey Wu, Ken Hong

Contact: ycm@stanford.edu

Award nominations: Spotlight talk at the ICLR-21 Workshop on Responsible AI

Keywords: deepfake, deepfakes, robustness, corruption, low-bandwidth, faceforensics


We look forward to seeing you at ICLR 2021!

Read More