Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Yingyu Liang

Yingyu Liang

Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles


Jun 29, 2021
Jiefeng Chen, Frederick Liu, Besim Avci, Xi Wu, Yingyu Liang, Somesh Jha

Figure 1 for Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles
Figure 2 for Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles
Figure 3 for Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles
Figure 4 for Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles

  Access Paper or Ask Questions

Towards Adversarial Robustness via Transductive Learning


Jun 15, 2021
Jiefeng Chen, Yang Guo, Xi Wu, Tianqi Li, Qicheng Lao, Yingyu Liang, Somesh Jha

Figure 1 for Towards Adversarial Robustness via Transductive Learning
Figure 2 for Towards Adversarial Robustness via Transductive Learning
Figure 3 for Towards Adversarial Robustness via Transductive Learning
Figure 4 for Towards Adversarial Robustness via Transductive Learning

  Access Paper or Ask Questions

Deep Online Fused Video Stabilization


Feb 02, 2021
Zhenmei Shi, Fuhao Shi, Wei-Sheng Lai, Chia-Kai Liang, Yingyu Liang

Figure 1 for Deep Online Fused Video Stabilization
Figure 2 for Deep Online Fused Video Stabilization
Figure 3 for Deep Online Fused Video Stabilization
Figure 4 for Deep Online Fused Video Stabilization
* 9 pages. Project page: https://zhmeishi.github.io/dvs/ 
  Access Paper or Ask Questions

PBoS: Probabilistic Bag-of-Subwords for Generalizing Word Embedding


Oct 21, 2020
Zhao Jinman, Shawn Zhong, Xiaomin Zhang, Yingyu Liang

Figure 1 for PBoS: Probabilistic Bag-of-Subwords for Generalizing Word Embedding
Figure 2 for PBoS: Probabilistic Bag-of-Subwords for Generalizing Word Embedding
Figure 3 for PBoS: Probabilistic Bag-of-Subwords for Generalizing Word Embedding
Figure 4 for PBoS: Probabilistic Bag-of-Subwords for Generalizing Word Embedding
* 16 pages including 4 pages of appendix. Accepted to Findings of EMNLP 2020 and SustaiNLP 2020. Code can be found at [https://github.com/jmzhao/pbos] 
  Access Paper or Ask Questions

Functional Regularization for Representation Learning: A Unified Theoretical Perspective


Aug 06, 2020
Siddhant Garg, Yingyu Liang

Figure 1 for Functional Regularization for Representation Learning: A Unified Theoretical Perspective
Figure 2 for Functional Regularization for Representation Learning: A Unified Theoretical Perspective
Figure 3 for Functional Regularization for Representation Learning: A Unified Theoretical Perspective
Figure 4 for Functional Regularization for Representation Learning: A Unified Theoretical Perspective

  Access Paper or Ask Questions

Can Adversarial Weight Perturbations Inject Neural Backdoors?


Aug 04, 2020
Siddhant Garg, Adarsh Kumar, Vibhor Goel, Yingyu Liang

Figure 1 for Can Adversarial Weight Perturbations Inject Neural Backdoors?
Figure 2 for Can Adversarial Weight Perturbations Inject Neural Backdoors?
Figure 3 for Can Adversarial Weight Perturbations Inject Neural Backdoors?
* Accepted as a conference paper at CIKM 2020 
  Access Paper or Ask Questions

Learning Entangled Single-Sample Gaussians in the Subset-of-Signals Model


Jul 10, 2020
Yingyu Liang, Hui Yuan

Figure 1 for Learning Entangled Single-Sample Gaussians in the Subset-of-Signals Model
* Appear in COLT'2020. Updates: corrected comments on existing works; added comparison to median estimator 
  Access Paper or Ask Questions

Robust Out-of-distribution Detection via Informative Outlier Mining


Jun 26, 2020
Jiefeng Chen, Yixuan Li, Xi Wu, Yingyu Liang, Somesh Jha

Figure 1 for Robust Out-of-distribution Detection via Informative Outlier Mining
Figure 2 for Robust Out-of-distribution Detection via Informative Outlier Mining
Figure 3 for Robust Out-of-distribution Detection via Informative Outlier Mining
Figure 4 for Robust Out-of-distribution Detection via Informative Outlier Mining

  Access Paper or Ask Questions

Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation


Apr 22, 2020
Xi Wu, Yang Guo, Jiefeng Chen, Yingyu Liang, Somesh Jha, Prasad Chalasani

Figure 1 for Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation
Figure 2 for Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation
Figure 3 for Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation
Figure 4 for Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation
* 25 pages, 6 figures 
  Access Paper or Ask Questions

Learning Entangled Single-Sample Distributions via Iterative Trimming


Apr 20, 2020
Hui Yuan, Yingyu Liang

Figure 1 for Learning Entangled Single-Sample Distributions via Iterative Trimming
Figure 2 for Learning Entangled Single-Sample Distributions via Iterative Trimming
Figure 3 for Learning Entangled Single-Sample Distributions via Iterative Trimming
* accepted in AISTAT 2020 
  Access Paper or Ask Questions

Gradients as Features for Deep Representation Learning


Apr 12, 2020
Fangzhou Mu, Yingyu Liang, Yin Li

Figure 1 for Gradients as Features for Deep Representation Learning
Figure 2 for Gradients as Features for Deep Representation Learning
Figure 3 for Gradients as Features for Deep Representation Learning
Figure 4 for Gradients as Features for Deep Representation Learning
* ICLR 2020 conference paper 
  Access Paper or Ask Questions

SimpleTran: Transferring Pre-Trained Sentence Embeddings for Low Resource Text Classification


Apr 10, 2020
Siddhant Garg, Rohit Kumar Sharma, Yingyu Liang

Figure 1 for SimpleTran: Transferring Pre-Trained Sentence Embeddings for Low Resource Text Classification
Figure 2 for SimpleTran: Transferring Pre-Trained Sentence Embeddings for Low Resource Text Classification
Figure 3 for SimpleTran: Transferring Pre-Trained Sentence Embeddings for Low Resource Text Classification
Figure 4 for SimpleTran: Transferring Pre-Trained Sentence Embeddings for Low Resource Text Classification

  Access Paper or Ask Questions

Robust Out-of-distribution Detection for Neural Networks


Apr 05, 2020
Jiefeng Chen, Yixuan Li, Xi Wu, Yingyu Liang, Somesh Jha

Figure 1 for Robust Out-of-distribution Detection for Neural Networks
Figure 2 for Robust Out-of-distribution Detection for Neural Networks
Figure 3 for Robust Out-of-distribution Detection for Neural Networks
Figure 4 for Robust Out-of-distribution Detection for Neural Networks

  Access Paper or Ask Questions

Robust Out-of-distribution Detection in Neural Networks


Mar 24, 2020
Jiefeng Chen, Yixuan Li, Xi Wu, Yingyu Liang, Somesh Jha

Figure 1 for Robust Out-of-distribution Detection in Neural Networks
Figure 2 for Robust Out-of-distribution Detection in Neural Networks
Figure 3 for Robust Out-of-distribution Detection in Neural Networks
Figure 4 for Robust Out-of-distribution Detection in Neural Networks

  Access Paper or Ask Questions

Sketching Transformed Matrices with Applications to Natural Language Processing


Feb 23, 2020
Yingyu Liang, Zhao Song, Mengdi Wang, Lin F. Yang, Xin Yang

Figure 1 for Sketching Transformed Matrices with Applications to Natural Language Processing
Figure 2 for Sketching Transformed Matrices with Applications to Natural Language Processing
Figure 3 for Sketching Transformed Matrices with Applications to Natural Language Processing
Figure 4 for Sketching Transformed Matrices with Applications to Natural Language Processing
* AISTATS 2020 
  Access Paper or Ask Questions

Learning Relationships between Text, Audio, and Video via Deep Canonical Correlation for Multimodal Language Analysis


Nov 30, 2019
Zhongkai Sun, Prathusha Sarma, William Sethares, Yingyu Liang

Figure 1 for Learning Relationships between Text, Audio, and Video via Deep Canonical Correlation for Multimodal Language Analysis
Figure 2 for Learning Relationships between Text, Audio, and Video via Deep Canonical Correlation for Multimodal Language Analysis
Figure 3 for Learning Relationships between Text, Audio, and Video via Deep Canonical Correlation for Multimodal Language Analysis
Figure 4 for Learning Relationships between Text, Audio, and Video via Deep Canonical Correlation for Multimodal Language Analysis

  Access Paper or Ask Questions

Shallow Domain Adaptive Embeddings for Sentiment Analysis


Aug 16, 2019
Prathusha K Sarma, Yingyu Liang, William A Sethares

Figure 1 for Shallow Domain Adaptive Embeddings for Sentiment Analysis
Figure 2 for Shallow Domain Adaptive Embeddings for Sentiment Analysis
Figure 3 for Shallow Domain Adaptive Embeddings for Sentiment Analysis
Figure 4 for Shallow Domain Adaptive Embeddings for Sentiment Analysis

  Access Paper or Ask Questions

Robust Attribution Regularization


May 23, 2019
Jiefeng Chen, Xi Wu, Vaibhav Rastogi, Yingyu Liang, Somesh Jha

Figure 1 for Robust Attribution Regularization
Figure 2 for Robust Attribution Regularization
Figure 3 for Robust Attribution Regularization
Figure 4 for Robust Attribution Regularization

  Access Paper or Ask Questions

Learning and Generalization in Overparameterized Neural Networks, Going Beyond Two Layers


Nov 12, 2018
Zeyuan Allen-Zhu, Yuanzhi Li, Yingyu Liang

Figure 1 for Learning and Generalization in Overparameterized Neural Networks, Going Beyond Two Layers
Figure 2 for Learning and Generalization in Overparameterized Neural Networks, Going Beyond Two Layers
Figure 3 for Learning and Generalization in Overparameterized Neural Networks, Going Beyond Two Layers

  Access Paper or Ask Questions

Recovery Guarantees for Quadratic Tensors with Limited Observations


Oct 31, 2018
Hongyang Zhang, Vatsal Sharan, Moses Charikar, Yingyu Liang

Figure 1 for Recovery Guarantees for Quadratic Tensors with Limited Observations
Figure 2 for Recovery Guarantees for Quadratic Tensors with Limited Observations
Figure 3 for Recovery Guarantees for Quadratic Tensors with Limited Observations
Figure 4 for Recovery Guarantees for Quadratic Tensors with Limited Observations

  Access Paper or Ask Questions

Generalizing Word Embeddings using Bag of Subwords


Sep 12, 2018
Jinman Zhao, Sidharth Mudgal, Yingyu Liang

Figure 1 for Generalizing Word Embeddings using Bag of Subwords
Figure 2 for Generalizing Word Embeddings using Bag of Subwords
Figure 3 for Generalizing Word Embeddings using Bag of Subwords
* Accepted to EMNLP 2018 
  Access Paper or Ask Questions

Learning Overparameterized Neural Networks via Stochastic Gradient Descent on Structured Data


Aug 20, 2018
Yuanzhi Li, Yingyu Liang

Figure 1 for Learning Overparameterized Neural Networks via Stochastic Gradient Descent on Structured Data
Figure 2 for Learning Overparameterized Neural Networks via Stochastic Gradient Descent on Structured Data
Figure 3 for Learning Overparameterized Neural Networks via Stochastic Gradient Descent on Structured Data
Figure 4 for Learning Overparameterized Neural Networks via Stochastic Gradient Descent on Structured Data

  Access Paper or Ask Questions

Linear Algebraic Structure of Word Senses, with Applications to Polysemy


Jul 20, 2018
Sanjeev Arora, Yuanzhi Li, Yingyu Liang, Tengyu Ma, Andrej Risteski

* Appear in the Transactions of the Association for Computational Linguistics 2018, link: https://transacl.org/ojs/index.php/tacl/article/view/1346 
  Access Paper or Ask Questions

N-Gram Graph, A Novel Molecule Representation


Jun 24, 2018
Shengchao Liu, Thevaa Chandereng, Yingyu Liang

Figure 1 for N-Gram Graph, A Novel Molecule Representation
Figure 2 for N-Gram Graph, A Novel Molecule Representation
Figure 3 for N-Gram Graph, A Novel Molecule Representation
Figure 4 for N-Gram Graph, A Novel Molecule Representation

  Access Paper or Ask Questions

Learning Mixtures of Linear Regressions with Nearly Optimal Complexity


Jun 12, 2018
Yuanzhi Li, Yingyu Liang

Figure 1 for Learning Mixtures of Linear Regressions with Nearly Optimal Complexity

  Access Paper or Ask Questions

Improving Adversarial Robustness by Data-Specific Discretization


May 25, 2018
Jiefeng Chen, Xi Wu, Yingyu Liang, Somesh Jha

Figure 1 for Improving Adversarial Robustness by Data-Specific Discretization
Figure 2 for Improving Adversarial Robustness by Data-Specific Discretization
Figure 3 for Improving Adversarial Robustness by Data-Specific Discretization
Figure 4 for Improving Adversarial Robustness by Data-Specific Discretization

  Access Paper or Ask Questions

A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors


May 14, 2018
Mikhail Khodak, Nikunj Saunshi, Yingyu Liang, Tengyu Ma, Brandon Stewart, Sanjeev Arora

Figure 1 for A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors
Figure 2 for A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors
Figure 3 for A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors
Figure 4 for A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors
* 11 pages, 2 figures, To appear in ACL 2018 
  Access Paper or Ask Questions

Matrix Completion and Related Problems via Strong Duality


Apr 25, 2018
Maria-Florina Balcan, Yingyu Liang, David P. Woodruff, Hongyang Zhang

Figure 1 for Matrix Completion and Related Problems via Strong Duality
Figure 2 for Matrix Completion and Related Problems via Strong Duality
Figure 3 for Matrix Completion and Related Problems via Strong Duality
Figure 4 for Matrix Completion and Related Problems via Strong Duality
* 37 pages, 4 figures 
  Access Paper or Ask Questions

Generalization and Equilibrium in Generative Adversarial Nets (GANs)


Aug 01, 2017
Sanjeev Arora, Rong Ge, Yingyu Liang, Tengyu Ma, Yi Zhang

Figure 1 for Generalization and Equilibrium in Generative Adversarial Nets (GANs)
Figure 2 for Generalization and Equilibrium in Generative Adversarial Nets (GANs)
Figure 3 for Generalization and Equilibrium in Generative Adversarial Nets (GANs)
Figure 4 for Generalization and Equilibrium in Generative Adversarial Nets (GANs)
* This is an updated version of an ICML'17 paper with the same title. The main difference is that in the ICML'17 version the pure equilibrium result was only proved for Wasserstein GAN. In the current version the result applies to most reasonable training objectives. In particular, Theorem 4.3 now applies to both original GAN and Wasserstein GAN 
  Access Paper or Ask Questions