Publications

(See the full publication list at Google scholar)

Selected Conference Publications

* represents equal contribution

2025

• Training Nonlinear Transformers for Chain-of-Thought Inference: A Theoretical Generalization Analysis
  Hongkang Li, Songtao Lu, Xiaodong Cui, Pin-Yu Chen, Meng Wang
  International Conference on Learning Representations (ICLR), Singapore, Apr. 2025.

• DUET: Decentralized Bilevel Optimization without Lower-Level Strong Convexity
  Zhen Qin, Zhuqing Liu, Songtao Lu, Yingbin Liang, Jia Liu
  International Conference on Learning Representations (ICLR), Singapore, Apr. 2025.

• Q-function Decomposition with Intervention Semantics for Factored Action Spaces
  Junkyu Lee, Tian Gao, Elliot Nelson, Miao Liu, Debarun Bhattacharjya, Songtao Lu
  International Conference on Artificial Intelligence and Statistics (AISTATS), Thailand, May 2025.

• Epigraph Based Multilevel Optimization (EMO) for Enhancing Chain-of-Thought Reasoning Capabilities
  Songtao Lu, Yanna Ding, Lior Horesh, Jianxi Gao, Malik Magdon-Ismail
  IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Kothaguda, India, Apr. 2025.

2024

• SPARKLE: A Unified Single-Loop Primal-Dual Framework for Decentralized Bilevel Optimization
  Shuchen Zhu, Boao Kong, Songtao Lu, Xinmeng Huang, Kun Yuan
  The Thirty-Eighth Annual Conference on Neural Information Processing Systems (NeurIPS), Vancouver, Canada, 2024.

• Federated Neuro-Symbolic Learning
  Pengwei Xing, Songtao Lu, Han Yu
  International Conference on Machine Learning (ICML), Vienna, Austria, 2024.

• How Do Nonlinear Transformers Learn and Generalize in In-Context Learning?
  Hongkang Li, Meng Wang, Songtao Lu, Xiaodong Cui, Pin-Yu Chen
  International Conference on Machine Learning (ICML), Vienna, Austria, 2024.

• FADAS: Towards Federated Adaptive Asynchronous Optimization
  Yujia Wang, Shiqiang Wang, Songtao Lu, Jinghui Chen
  International Conference on Machine Learning (ICML), Vienna, Austria, 2024.

• Distributed Bilevel Optimization with Communication Compression
  Yutong He, Jie Hu, Xinmeng Huang, Songtao Lu, Bin Wang, Kun Yuan
  International Conference on Machine Learning (ICML), Vienna, Austria, 2024.

• SF-DQN: Provable Knowledge Transfer using Successor Feature for Deep Reinforcement Learning
  Shuai Zhang, Heshan Devaka Fernando, Miao Liu, Keerthiram Murugesan, Songtao Lu, Pin-Yu Chen, Tianyi Chen, Meng Wang
  International Conference on Machine Learning (ICML), Vienna, Austria, 2024.

• Toward Byzantine-Robust Decentralized Federated Learning
  Minghong Fang, Zifan Zhang, Hairi, Prashant Khanduri, Jia Liu, Songtao Lu, Yuchen Liu, Neil Zhenqiang Gong.
  The ACM Conference on Computer and Communications Security (CCS), Salt Lake City, 2024.

• PILOT: An O(1/K)-Convergent Approach for Policy Evaluation with Nonlinear Function Approximation
  Zhuqing Liu, Xin Zhang, Jia Liu, Zhengyuan Zhu, Songtao Lu
  International Conference on Learning Representations (ICLR), 2024.
  - Spotlight Presentation: 5%

2023

• SLM: A Smoothed First-Order Lagrangian Method for Structured Constrained Nonconvex Optimization
  Songtao Lu
  The Thirty-Seventh Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, 2023.

• A Generalized Alternating Method for Bilevel Optimization under the Polyak-Łojasiewicz Condition
  Quan Xiao, Songtao Lu, Tianyi Chen
  The Thirty-Seventh Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, 2023.

• On the Convergence and Sample Complexity Analysis of Deep Q-Networks with ε-Greedy Exploration
  Shuai Zhang, Hongkang Li, Meng Wang, Miao Liu, Pin-Yu Chen, Songtao Lu, Sijia Liu, Keerthiram Murugesan, Subhajit Chaudhury
  The Thirty-Seventh Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, 2023.

• Bilevel Optimization with Coupled Decision-Dependent Distributions
  Songtao Lu
  International Conference on Machine Learning (ICML), Honolulu, Hawai'i, 2023.

• Prometheus: Taming Sample and Communication Complexities in Constrained Decentralized Stochastic Bilevel Learning
  Zhuqing Liu, Xin Zhang, Prashant Khanduri, Songtao Lu, Jia Liu
  International Conference on Machine Learning (ICML), Honolulu, Hawai'i, 2023.

• Compressed Decentralized Proximal Stochastic Gradient Method for Nonconvex Composite Problems with Heterogeneous Data
  Yonggui Yan, Jie Chen, Pin-Yu Chen, Xiaodong Cui, Songtao Lu, Yangyang Xu
  International Conference on Machine Learning (ICML), Honolulu, Hawai'i, 2023.

• Meta-DAG: Meta Causal Discovery via Bilevel Optimization
  Songtao Lu*, Tian Gao*
  IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Rhodes Island, Greece, 2023.

• Min-Max Multi-Objective Bilevel Optimization with Applications in Robust Machine Learning
  Alex Gu, Songtao Lu, Parikshit Ram, Tsui-Wei Weng
  The Eleventh International Conference on Learning Representations (ICLR), Kigali Rwanda, 2023.

• Joint Edge-Model Sparse Learning is Provably Efficient for Graph Neural Networks
  Shuai Zhang, Meng Wang, Pin-Yu Chen, Sijia Liu, Songtao Lu, Miao Liu
  The Eleventh International Conference on Learning Representations (ICLR), Kigali Rwanda, 2023.

• Distributed Offline Policy Optimization Over Batch Data
  Han Shen, Songtao Lu, Xiaodong Cui, Tianyi Chen
  The 26th International Conference on Artificial Intelligence and Statistics (AISTATS), Palau de Congressos, Valencia, Spain, 2023.

2022

• A Stochastic Linearized Augmented Lagrangian Method for Decentralized Bilevel Optimization
  Songtao Lu, Siliang Zeng, Xiaodong Cui, Mark S. Squillante, Lior Horesh, Brian Kingsbury, Jia Liu, Mingyi Hong
  The Thirty-Sixth Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, 2022.

• Understanding Benign Overfitting in Gradient-based Meta Learning
  Lisha Chen, Songtao Lu, Tianyi Chen
  The Thirty-Sixth Conference on Neural Information Processing Systems (NeurIPS), New Orleans, LA, 2022.

• A Single-Loop Gradient Descent and Perturbed Ascent Algorithm for Nonconvex Functional Constrained Optimization
  Songtao Lu
  The 39th International Conference on Machine Learning (ICML), Baltimore, Maryland, 2022.

• Distributed Adversarial Training to Robustify Deep Neural Networks at Scale
  Gaoyuan Zhang*, Songtao Lu*, Yihua Zhang, Xiangyi Chen, Pin-Yu Chen, Quanfu Fan, Lee Martie, Lior Horesh, Mingyi Hong, Sijia Liu
  The 38th Conference on Uncertainty in Artificial Intelligence (UAI), Eindhoven, Netherlands, 2022.
  - Oral Presentation: 5% [36/712] and Best Paper Runner-Up Award

• Learning to Generate Image Source-Agnostic Universal Adversarial Perturbations
  Pu Zhao, Parikshit Ram, Songtao Lu, Yuguang Yao, Djallel Bouneffouf, Xue Lin, Sijia Liu
  The 31st International Joint Conference on Artificial Intelligence (IJCAI), Messe Wien, Vienna, Austria, 2022.

• Decentralized Bilevel Optimization for Personalized Client Learning
  Songtao Lu*, Xiaodong Cui*, Mark S. Squillante, Brain Kingsbury, and Lior Horesh
  The 47th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Singapore, 2022.

• Finite-Time Convergence and Sample Complexity of Multi-Agent Actor-Critic Reinforcement Learning with Average Reward
  FNU Hairi, Jia Liu, Songtao Lu
  International Conference on Learning Representations (ICLR), 2022.
  - Spotlight Presentation: 5.19% [176/3391]

• Understanding Latent Correlation-based Multiview Learning and Self-Supervision: An Identifiability Perspective
  Qi Lyu, Xiao Fu, Weiran Wang, Songtao Lu
  International Conference on Learning Representations (ICLR), 2022.
  - Spotlight Presentation: 5.19% [176/3391]

• Adversarial Examples Can Be Effective Data Augmentation for Unsupervised Machine Learning
  Chia-Yi Hsu, Pin-Yu Chen, Songtao Lu, Sijia Liu, Chia-Mu Yu
  The 36th AAAI Conference on Artificial Intelligence (AAAI), Vancouver, Canada, 2022.

• Zeroth-Order Optimization for Composite Problems with Functional Constraints
  Zichong Li, Pin-Yu Chen, Sijia Liu, Songtao Lu, Yangyang Xu
  The 36th AAAI Conference on Artificial Intelligence (AAAI), Vancouver, Canada, 2022.
  - Oral Presentation: 3.75%

2021

• Taming Communication and Sample Complexities in Decentralized Policy Evaluation for Cooperative Multi-Agent Reinforcement Learning
  Xin Zhang, Zhuqing Liu, Jia Liu, Zhengyuan Zhu, Songtao Lu
  The Thirty-Fifth Conference on Neural Information Processing Systems (NeurIPS), 2021.

• Decentralized Policy Gradient Descent Ascent for Safe Multi-Agent Reinforcement Learning
  Songtao Lu, Kaiqing Zhang, Tianyi Chen, Tamer Basar, and Lior Horesh
  The Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI), vol. 35, no. 10, pp. 8767-8775, 2021. (Supplemental Materials)

• Rate-Improved Inexact Augmented Lagrangian Method for Constrained Nonconvex Optimization
  Zichong Li, Pin-Yu Chen*, Sijia Liu*, Songtao Lu*, Yangyang Xu*
  The 24th International Conference on Artificial Intelligence and Statistics (AISTATS), 2021.

• Training Logical Neural Networks by Primal-Dual Methods for Neuro-Symbolic Reasoning
  Songtao Lu*, Naweed Khan*, Ismail Akhalwaya*, Ryan Riegel, Lior Horesh, Alexander Gray
  The 46th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Toronto, Ontario, Canada, 2021.

• Federated Acoustic Modeling for Automatic Speech Recognition
  Xiaodong Cui, Songtao Lu, Brian Kingsbury
  The 46th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Toronto, Ontario, Canada, 2021.

• On the Convergence of Randomized Bregman Coordinate Descent for Non-Lipschitz Composite Problems
  Tianxiang Gao, Songtao Lu, Jia Liu, Chris Chu
  The 46th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Toronto, Ontario, Canada, 2021.

2020

• Finding Second-Order Stationary Points Efficiently in Smooth Nonconvex Linearly Constrained Optimization Problems
  Songtao Lu, Meisam Razaviyayn, Bo Yang, Kejun Huang, Mingyi Hong
  The Thirty-Fourth Conference on Neural Information Processing Systems (NeurIPS), 2020.
  - Spotlight Presentation: 3% [280/9454] (Code Available)

• Decentralized TD Tracking with Linear Function Approximation and its Finite-Time Analysis
  Gang Wang, Songtao Lu, Georgios B. Giannakis, Gerald Tesauro, Jian Sun
  The Thirty-Fourth Conference on Neural Information Processing Systems (NeurIPS), 2020.

• ScaleCom: Scalable Sparsified Gradient Compression for Communication-Efficient Distributed Training
  Chia-Yu Chen, Jiamin Ni, Songtao Lu, Xiaodong Cui, Pin-Yu Chen, Xiao Sun, Naigang Wang, Swagath Venkataramani, Vijayalakshmi Srinivasan, Wei Zhang, Kailash Gopalakrishnan
  The Thirty-Fourth Conference on Neural Information Processing Systems (NeurIPS), 2020.

• Improving the Sample and Communication Complexity for Decentralized Non-Convex Optimization: Joint Gradient Estimation and Tracking
  Haoran Sun*, Songtao Lu*, Mingyi Hong*
  The 37th International Conference on Machine Learning (ICML), 2020.

• Min-Max Optimization without Gradients: Convergence and Applications to Black-Box Evasion and Poisoning Attacks
  Sijia Liu*, Songtao Lu*, Xiangyi Chen*, Yao Feng*, Kaidi Xu*, Abdullah Al-Dujaili*, Minyi Hong, Una-May Obelilly
  The 37th International Conference on Machine Learning (ICML), 2020.

• Decentralized Stochastic Non-Convex Optimization over Weakly Connected Time-Varying Digraphs
  Songtao Lu*, Chai Wah Wu*
  The 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Barcelona, Spain, 2020.

• No-Regret Non-Convex Online Meta-Learning
  Zhenxun Zhuang, Yunlong Wang, Kezi Yu, Songtao Lu
  The 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Barcelona, Spain, 2020.

• Decentralized Federated Learning for Electronic Health Records
  Songtao Lu, Yawen Zhang, Yunlong Wang
  The 54th Annual Conference on Information Sciences and Systems (CISS), Princenton, 2020.

2019

• PA-GD: On the Convergence of Perturbed Alternating Gradient Descent to Second-Order Stationary Points for Structured Nonconvex Optimization
  Songtao Lu, Mingyi Hong, Zhengdao Wang
  The 36th International Conference on Machine Learning (ICML), Long Beach, CA, 2019.
  - Long Oral Talk: 4.6% and ICML Travel Award

• Block Alternating Optimization for Non-convex Min-max Problems: Algorithms and Applications in Signal Processing and Communications
  Songtao Lu, Ioannis Tsaknakis, Mingyi Hong
  The 44th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Brighton, UK, 2019.

• Perturbed Projected Gradient Descent Converges to Approximate Second-order Points for Bound Constrained Nonconvex Problems
  Songtao Lu*, Ziping Zhao*, Kejun Huang, and Mingyi Hong
  The 44th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Brighton, UK, 2019.

• Fast and Global Optimal Nonconvex Matrix Factorization via Perturbed Alternating Proximal Point
  Songtao Lu, Mingyi Hong, Zhengdao Wang
  The 44th IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Brighton, UK, 2019.

• GNSD: a Gradient-Tracking Based Nonconvex Stochastic Algorithm for Decentralized Optimization
  Songtao Lu*, Xinwei Zhang*, Haoran Sun*, Mingyi Hong
  IEEE Data Science Workshop (DSW), Minneapolis, MN, 2019.

• Alternating Gradient Descent Ascent for Nonconvex Min-Max Problems in Robust Learning and GANs
  Songtao Lu, Rahul Singh, Xiangyi Chen, Yongxin Chen, Mingyi Hong
  The 53rd Asilomar Conference on Signals, Systems, and Computers (Asilomar), Pacific Grove, CA, 2019.

2018 and earlier

• Distributed optimization for Generalized Phase Retrieval Over Networks
  Ziping Zhao, Songtao Lu, Mingyi Hong, Daniel P. Palomar
  The 52nd Asilomar Conference on Signals, Systems, and Computers (Asilomar), Pacific Grove, CA, 2018.

• A Stochastic Nonconvex Splitting Method for Symmetric Nonnegative Matrix Factorization
  Songtao Lu, Mingyi Hong, Zhengdao Wang
  The 20th International Conference on Artificial Intelligence and Statistics (AISTATS), Fort Lauderdale, Florida, 2017.
  - AISTATS Travel Award

• Accelerated Algorithms for Eigen-Value Decomposition with Application to Spectral Clustering
  Songtao Lu, Zhengdao Wang
  The 49th Annual Asilomar Conference on Signals, Systems, and Computers (Asilomar), Pacific Grove, CA, 2015.

• Achievable Rates of Uplink Multiuser Massive MIMO Systems with Estimated Channels
  Songtao Lu, Zhengdao Wang
  IEEE Global Communications Conference (GLOBECOM), Austin, TX, 2014.

Book Chapter

• Graph-Aware Federated Learning
  Songtao Lu, Pengwei Xing, Yu Han
  Federated Learning Theory and Practice, pp. 181 - 197, Elsevier, 2024.

Selected Journal Publications

* represents corresponding author

• Bilevel Joint Unsupervised and Supervised Training for Automatic Speech Recognition
  Xiaodong Cui, AFM Saif, Songtao Lu, Lisha Chen, Tianyi Chen, Brian Kingsbury, George Saon
  IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2024. (Accepted)

• Graph Neural Networks with Adaptive Structures
  Zepeng Zhang, Songtao Lu, Zengfeng Huang, Ziping Zhao
  IEEE Journal of Selected Topics in Signal Processing, 2024.

• Stochastic Inexact Augmented Lagrangian Method for Nonconvex Expectation Constrained Optimization
  Zichong Li, Pin-Yu Chen, Sijia Liu, Songtao Lu, Yangyang Xu
  Computational Optimization and Applications, vol. 87, pp. 117 - 147, 2024.

• BiG-Fed: Bilevel Optimization Enhanced Graph-Aided Federated Learning
  Pengwei Xing, Songtao Lu, Lingfei Wu, Han Yu
  IEEE Transactions on Big Data, vol. 10, no. 6, pp. 903 - 914, 2024.

• An Efficient Learning Framework for Federated XGBoost Using Secret Sharing and Distributed Optimization
  Lunchen Xie, Jiaqi Liu, Songtao Lu, Tsung-Hui Chang, Qingjiang Shi
  ACM Transactions on Intelligent Systems and Technology, vol. 13, issue 5, 2022.

• Linearized ADMM Converges to Second-Order Stationary Points for Non-Convex Problems
  Songtao Lu, Jason Lee, Meisam Razaviyayn, Mingyi Hong
  IEEE Transactions on Signal Processing, vol. 69, pp. 4859 - 4874, 2021.

• Spectrum Truncation Power Iteration for Agnostic Matrix Phase Retrieval
  Lewis Liu, Songtao Lu*, Tuo Zhao, Zhaoran Wang
  IEEE Transactions on Signal Processing, vol. 69, pp. 3991 - 4006, 2021

• Non-Convex Min-Max Optimization: Applications, Challenges, and Recent Theoretical Advances
  Meisam Razaviyayn, Tianjian Huang, Songtao Lu, Maher Nouiehed, Maziar Sanjabi, Mingyi Hong
  IEEE Signal Processing Magazine, vol. 37, issue 5, pp. 55 - 66, 2020.

• Distributed Learning in the Non-Convex World: From Batch to Streaming Data, and Beyond
  Tsung-Hui Chang, Mingyi Hong, Hoi-To Wai, Xinwei Zhang, Songtao Lu
  IEEE Signal Processing Magazine, vol. 37 , issue 3, pp. 26 - 38, 2020.

• Hybrid Block Successive Approximation for One-Sided Non-Convex Min-Max Problems: Algorithms and Applications
  Songtao Lu, Ioannis Tsaknakis, Mingyi Hong, Yongxin Chen
  IEEE Transactions on Signal Processing, vol. 68, pp. 3676 - 3691, 2020.

• Training Optimization and Performance of Single Cell Uplink System with Massive-Antennas Base Station
  Songtao Lu, Zhengdao Wang
  IEEE Transactions on Communications, vol. 67, no. 2, pp. 1570 - 1585, 2019.

• Spatial Transmitter Density Allocation for Frequency-Selective Wireless Ad Hoc Networks
  Songtao Lu, Zhengdao Wang
  IEEE Transactions on Wireless Communications, vol. 18, no. 1, pp. 473 - 486, 2019.

• A Nonconvex Splitting Method for Symmetric Nonnegative Matrix Factorization: Convergence Analysis and Optimality
  Songtao Lu, Mingyi Hong, Zhengdao Wang
  IEEE Transactions on Signal Processing, vol. 65, no. 12, pp. 3120 - 3135, 2017.

• Inexact Block Coordinate Descent Methods for Symmetric Nonnegative Matrix Factorization
  Qingjiang Shi, Haoran Sun, Songtao Lu, Mingyi Hong, Meisam Razaviyayn
  IEEE Transactions on Signal Processing, vol. 65, no. 22, pp. 5995 - 6008, 2017.

• Throughput of Underwater Wireless Ad Hoc Networks with Random Access: A Physical Layer Perspective
  Songtao Lu, Zhengdao Wang, Zhaohui Wang, Shengli Zhou
  IEEE Transactions on Wireless Communications, vol. 14, no. 11, pp. 6257 - 6268, 2015.

Selected Workshop Papers

• How Do Nonlinear Transformers Acquire Generalization-Guaranteed CoT Ability?
  Hongkang Li, Meng Wang, Songtao Lu, Xiaodong Cui, Pin-Yu Chen
  ICML Workshop on Theoretical Foundations of Foundation Models (TF2M), Vienna, Austria, July 27, 2024.

• Transformers as Multi-Task Feature Selectors: Generalization Analysis of In-Context Learning
  Hongkang Li, Meng Wang, Songtao Lu, Hui Wan, Xiaodong Cui, Pin-Yu Chen
  NeurIPS Workshop on Mathematics of Modern Machine Learning (M3L), New Orleans, LA, U.S.A., Dec. 16, 2023.

• Conditional Moment Alignment for Improved Generalization in Federated Learning
  Jayanth Reddy Regatti, Songtao Lu, Abhishek Gupta, Ness Shroff
  NeurIPS Workshop on Federated Learning (FL-NeurIPS), New Orleans, LA, USA, December 2, 2022.
  - Oral Presentation: 11% [12/103] and Outstanding Paper Award