Wang, Weiran (汪蔚然) Senior Research Scientist Google 1945 Charleston Rd Mountain View, CA 94043 Email: weiranwang@google.com Email: weiranwang@ttic.edu previous homepage

Biography

From 2019 to 2020, I was a research scientist at Salesforce Research. From 2017 to 2019, I was an applied scientist at Amazon Alexa. From 2014 to 2017, I was a postdoc at Toyota Technological Institute at Chicago working with Professor Karen Livescu and Professor Nathan Srebro. I obtained my Ph.D. from the EECS Department at UC Merced in 2013, under the supervision of Professor Miguel A. Carreira-Perpinan. Prior to that, I obtained Master's degree in Computer Science from Chengdu Institute of Computer Applications (Chengdu, China), Chinese Academy of Sciences in 2008, and my Bachelor's degree in Computer Science from Huazhong University of Science and Technology (Wuhan, China) in 2005.

Publications

• Weiran Wang, Ke Hu, and Tara Sainath. Deliberation of Streaming RNN-Transducer by Non-autoregressive Decoding. ICASSP, 2022.
  [arXiv version]

• Tara Sainath, Yanzhang He, Arun Narayanan, Rami Botros, Weiran Wang, David Qiu, Chung-cheng Chiu, Rohit Prabhavalkar, Alexander Gruenstein, Anmol Gulati, Bo Li, David Rybach, Emmanuel Guzman, Ian McGraw, James Qin, Krzysztof Choromanski, Qiao Liang, Robert David, Ruoming Pang, Shuoyiin Chang, Trevor Strohman, W. Ronny Huang, Wei Han, Yonghui Wu, and Yu Zhang. Improving the Latency and Quality of Cascaded Encoders. ICASSP, 2022.

• Qi Lyu, Xiao Fu, Weiran Wang, and Songtao Lu. Latent Correlation-Based Multiview Learning and Self-Supervision: A Unifying Perspective. ICLR, 2022.
  [arXiv version]

• Junwen Bai, Weiran Wang, and Carla Gomes. Contrastively Disentangled Sequential Variational Autoencoder. NeurIPS, 2021.
  [arXiv version] [implementation]

• Junwen Bai, Weiran Wang, Yingbo Zhou, and Caiming Xiong. Representation Learning for Sequence Data with Deep Autoencoding Predictive Components. International Conference on Learning Representations (ICLR), 2021.
  [arXiv version] [implementation]

• Weiran Wang, Guangsen Wang, Aadyot Bhatnagar, Yingbo Zhou, Caiming Xiong, and Richard Socher. An investigation of phone-based subword units for end-to-end speech recognition. Interspeech, 2020.
  For Switchboard, our phone-based BPE system achieves 6.8%/14.4% word error rate (WER) on the Switchboard/CallHome portion of the test set while joint decoding achieves 6.3%/13.3% WER.
  On Fisher + Switchboard, joint decoding leads to 4.9%/9.5% WER, setting new milestones for telephony speech recognition.
  [arXiv version] [implementation]

• Yang Chen, Weiran Wang, and Chao Wang. Semi-supervised ASR by End-to-end Self-training. Interspeech, 2020.
  [arXiv version]

• Weimin Wang, Weiran Wang, Ming Sun, and Chao Wang. Acoustic Scene Analysis with Multi-head Attention Networks. Interspeech, 2020.
  [arXiv version]

• Weiran Wang, Qingming Tang, and Karen Livescu. Unsupervised Pre-training of Bidirectional Speech Encoders via Masked Reconstruction. ICASSP, 2020.
  [arXiv version]

• Chieh-chi Kao, Ming Sun, Weiran Wang, and Chao Wang. A Comparison of Pooling Methods on LSTM Models for Rare Acoustic Event Classification. ICASSP, 2020.

• Sundar Harsha, Weiran Wang, Ming Sun and Chao Wang. Raw Waveform based End-to-end Deep Convolutional Network for Spatial Localization of Multiple Acoustic Sources. ICASSP, 2020.

• Chao Gao, Dan Garber, Nathan Srebro, Jialei Wang, Weiran Wang (by α-β order). Stochastic Canonical Correlation Analysis. Accepted by Journal of Machine Learning Research.
  [arXiv version]

• Gustavo Aguilar, Viktor Rozgic, Weiran Wang, and Chao Wang. Multimodal and Multi-view Models for Emotion Recognition. Association for Computational Linguistics (ACL), 2019.
  [arXiv version]

• Weiran Wang and Nathan Srebro. Stochastic Nonconvex Optimization with Large Minibatches. Algorithmic Learning Theory (ALT), 2019.
  [arXiv version]

• Weiran Wang, Chieh-chi Kao, and Chao Wang. A simple model for detection of rare sound events. Interspeech, 2018.
  

• Chieh-chi Kao, Weiran Wang, Ming Sun, and Chao Wang. R-CRNN: Region-based Convolutional Recurrent Neural Network for Audio Event Detection. Interspeech, 2018.
  

• Jialei Wang^*, Weiran Wang^*, Dan Garber, and Nathan Srebro. Efficient Coordinate-wise Leading Eigenvector Computation. Algorithmic Learning Theory (ALT), 2018.
  [arXiv version]

• Qingming Tang, Weiran Wang, and Karen Livescu. Acoustic feature learning using cross-domain articulatory measurements. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2018.
  [arXiv version]

• Taehwan Kim, Jonathan Keane, Weiran Wang, Hao Tang, Jason Riggle, Gregory Shakhnarovich, Diane Brentari, and Karen Livescu. Lexicon-Free Fingerspelling Recognition from Video: Data, Models, and Signer Adaptation. Accepted by Computer Speech and Language.
  [arxiv version]

• Qingming Tang, Weiran Wang, and Karen Livescu. Acoustic feature learning with deep variational canonical correlation analysis. Interspeech, 2017.

• Jialei Wang^*, Weiran Wang^*, and Nathan Srebro. Memory and Communication Efficient Distributed Stochastic Optimization with Minibatch Prox. Conference On Learning Theory (COLT), 2017.
  [arXiv version]

• Wanjia He, Weiran Wang, and Karen Livescu. Multi-view Recurrent Neural Acoustic Word Embeddings. International Conference on Learning Representations (ICLR), 2017.
  [arXiv version] [Tensorflow implementation]

• Hao Tang, Weiran Wang, Kevin Gimpel, and Karen Livescu. End-to-end Training Approaches for Discriminative Segmental Models. IEEE Spoken Language Technology Workshop (SLT), 2016
  [preprint]

• Weiran Wang^*, Jialei Wang^*, Dan Garber, and Nathan Srebro. Efficient Globally Convergent Stochastic Optimization for Canonical Correlation Analysis. Advances in Neural Information Processing Systems (NIPS), 2016.
  [arXiv version]

• Weiran Wang, Hao Tang, and Karen Livescu. Triphone State-tying via Deep Canonical Correlation Analysis. Interspeech, 2016.
  [paper preprint]

• Hao Tang, Weiran Wang, Kevin Gimpel, and Karen Livescu. Efficient Segmental Cascades for Speech Recognition. Interspeech, 2016.
  [arXiv version]

• Tomer Michaeli, Weiran Wang, and Karen Livescu. Nonparametric Canonical Correlation Analysis. International Conference on Machine Learning (ICML), 2016.
  [arXiv version]

• Weiran Wang and Karen Livescu. Large-scale Approximate Kernel Canonical Correlation Analysis. International Conference on Learning Representations (ICLR), 2016.
  [arXiv version] [Matlab implementation ]

• Herman Kamper, Weiran Wang, and Karen Livescu. Deep Convolutional Acoustic Word Embeddings using Word-pair Side Information. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016.
  [paper preprint]

• Taehwan Kim, Weiran Wang, Hao Tang, and Karen Livescu. Signer-independent Fingerspelling Recognition with Deep Neural Network Adaptation. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016. (ICASSP2016 Best Student Paper of Speech and Language Processing!)
  [paper preprint]

• Weiran Wang, Raman Arora, Nathan Srebro, and Karen Livescu. Stochastic Optimization for Deep CCA via Nonlinear Orthogonal Iterations. Annual Allerton Conference on Communication, Control, and Computing (ALLERTON), 2015.
  [paper preprint]

• Hao Tang, Weiran Wang, Kevin Gimpel, and Karen Livescu. Discriminative Segmental Cascades for Feature-Rich Phone Recognition. IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), 2015. (Best paper nomination!)
  [paper preprint]

• Weiran Wang, Raman Arora, Karen Livescu, and Jeff Bilmes. On Deep Multi-View Representation Learning. International Conference on Machine Learning (ICML), 2015.
  [XRMB dataset] [Tensorflow implementation!]

• Ang Lu, Weiran Wang, Mohit Bansal, Kevin Gimpel, and Karen Livescu. Deep Multilingual Correlation for Improved Word Embeddings. Conference of the North American Chapter of the Association for Computational Linguistics -- Human Language Technologies (NAACL-HLT), 2015.

• Weiran Wang, Raman Arora, Karen Livescu, and Jeff Bilmes. Unsupervised Learning of Acoustic Features via Deep Canonical Correlation Analysis. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2015.

• Weiran Wang, Raman Arora, and Karen Livescu. Reconstruction of Articulatory Measurements with Smoothed Low-rank Matrix Completion. IEEE Spoken Language Technology Workshop (SLT), 2014.
  

• Weiran Wang and Miguel A. Carreira-Perpinan. The Role of Dimensionality Reduction in Classification. AAAI Conference on Artificial Intelligence (AAAI), 2014.

• Miguel A. Carreira-Perpinan and Weiran Wang. LASS: A Simple Assignment Model with Laplacian Smoothing. AAAI Conference on Artificial Intelligence (AAAI), 2014.

• Miguel A. Carreira-Perpinan and Weiran Wang. Distributed Optimization of Deeply Nested Systems. International Conference on Artificial Intelligence and Statistics (AISTATS), 2014.

• Weiran Wang. Mean-shift algorithms for manifold denoising, matrix completion and clustering. PhD thesis, University of California, Merced.

• Weiran Wang and Miguel A. Carreira-Perpinan. Nonlinear Low-dimensional Regression using Auxiliary Coordinates. International Conference on Artificial Intelligence and Statistics (AISTATS), 2012.

• Weiran Wang, Miguel A. Carreira-Perpinan, and Zhengdong Lu. A Denoising View of Matrix Completion. Advances in Neural Information Processing Systems (NIPS), 2011.

• Weiran Wang and Miguel A. Carreira-Perpinan. Manifold Blurring Mean Shift Algorithms for Manifold Denoising. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2010.

Reports

• Yang Chen, Weiran Wang, I-fan Chen, and Chao Wang. Data Techniques for Online End-to-end Speech Recognition.
  [arXiv version]

• Weiran Wang, Jialei Wang, Mladen Kolar, and Nathan Srebro. Distributed Stochastic Multi-task Learning with Graph Regularization.
  [arXiv version]

• Weiran Wang, Xinchen Yan, Honglak Lee, and Karen Livescu. Deep Variational Canonical Correlation Analysis.
  [arXiv version]

• Weiran Wang. On Column Selection in Kernel Canonical Correlation Analysis. Feb 5, 2016, arXiv:1602.02172 [cs.LG].
  [arXiv version]

• Qingming Tang, Lifu Tu, Weiran Wang, and Jinbo Xu. Network Inference by Learned Node-Specific Degree Prior. Feb 7, 2016, arXiv:1602.02386 [stat.ML].
  [arXiv version]

• Weiran Wang, Raman Arora, Karen Livescu, and Jeff Bilmes. On Deep Multi-View Representation Learning: Objectives and Optimization. Feb 2, 2016, arXiv:1602.01024 [cs.LG].
  [arXiv version] [Matlab implementation ]

• Weiran Wang and Miguel A. Carreira-Perpinan: "The Laplacian K-modes algorithm for clustering" (2014). Unpublished manuscript, June 15, 2014, arXiv:1406.3895 [cs.LG].
  [external link]

• Weiran Wang and Miguel A. Carreira-Perpinan: "Projection onto the probability simplex: an efficient algorithm with a simple proof, and an application". Unpublished manuscript, September 3, 2013, arXiv:1309.1541 [cs.LG].
  [external link]

  I have got inquiries asking if different projections can be solved with the same "sorting" idea. Two solvable examples are given below.

• Weiran Wang and Canyi Lu: "Projection onto the capped simplex". March 3, 2015, arXiv:1503.01002 [cs.LG].
  [external link]

• Weiran Wang: "An O(nlogn) projection operator for weighted l1-norm regularization with sum constraint". March 2, 2015, arXiv:1503.00600 [cs.LG].
  [external link]

• Miguel A. Carreira-Perpinan and Weiran Wang: "The K-modes algorithm for clustering". Unpublished manuscript, April 23, 2013, arXiv:1304.6478 [cs.LG].
  [external link]