Songtao Lu (卢松涛)
Senior Research Scientist
Mathematics and Theoretical Computer Science Department
Thomas J. Watson Research Center & MIT-IBM Watson AI Lab
IBM Research
Yorktown Heights, New York 10598, USA
I received my Ph.D. from the Department of Electrical and Computer Engineering at Iowa State University in August 2018 under the supervision of Professor Mingyi Hong (now at UMN) and Professor Zhengdao Wang (now at GMU). I was a postdoc associate hosted by Prof. Hong with the Department of Electrical and Computer Engineering at the University of Minnesota Twin Cities from Sept. 2018 to Sept. 2019, then I was an AI resident with the Mathematics and Theoretical Computer Science Department at the Thomas J. Watson Research Center from Sept. 2019 to Aug. 2020.
Currently, I am a Senior Research Scientist at the Thomas J. Watson Research Center, an IBM Principal Investigator at the MIT-IBM Watson AI Lab, and an IBM PI of the RPI-IBM AI Research Collaboration program. Also, I serve as senior personnel at the NSF AI Institute for Future Edge Networks and Distributed Intelligence (AI-EDGE).
My research interests lie in the areas of artificial intelligence, machine learning, optimization, and signal processing.
Selected Publications
Songtao Lu
NeurIPS, 2023
Songtao Lu
ICML, 2023
Songtao Lu, Siliang Zeng, Xiaodong Cui, Mark S. Squillante, Lior Horesh, Brian Kingsbury, Jia Liu, Mingyi Hong
NeurIPS, 2022
Songtao Lu
ICML, 2022
Songtao Lu, Jason Lee, Meisam Razaviyayn, Mingyi Hong
IEEE Trans. Signal Process., 2021
Songtao Lu, Kaiqing Zhang, Tianyi Chen, Tamer Basar, Lior Horesh
AAAI, 2021
Songtao Lu, Meisam Razaviyayn, Bo Yang, Kejun Huang, Mingyi Hong
NeurIPS, 2020
Songtao Lu, Ioannis Tsaknakis, Mingyi Hong, Yongxin Chen
IEEE Trans. Signal Process., 2020
Songtao Lu, Mingyi Hong, Zhengdao Wang
ICML, 2019
News
01/16/2024 Our work on MinMax optimization for policy evaluation with nonlinear function approximation is selected as a spotlight presentation by ICLR 2024.
12/13/2023 I am thrilled to have received the IBM Research Accomplishment Award for my contributions to advancing optimization techniques for next-generation distributed intelligence.
12/13/2023 Four papers were accepted by ICASSP 2024.
12/11/2023-12/16/2023 Traveling to New Orleans!
11/07/2023 Two RPI-IBM projects are funded.
11/02/2023 It is my great pleasure to have received the IBM Plateau Invention Achievement Award.
09/21/2023 Three papers on constrained optimization, bilevel optimization, and Q-learning have been accepted by NeurIPS 2023.
08/21/2023 I have been elevated as an IEEE Senior Member.
08/18/2023 Our work on stochastic inexact augmented Lagrangian method for nonconvex expectation constrained optimization was accepted by Computational Optimization and Applications.
07/14/2023 Our work on decentralized constrained min-max optimization was accepted by ACM MobiHoc 2023.
07/03/2023 I am truly honored to receive the IBM Entrepreneur Award.
05/18/2023 I will serve as a Senior Area Chair for AAAI 2024.
04/24/2023 Three papers were accepted by ICML 2023.
03/08/2023 I will serve as an Area Chair for NeurIPS 2023.
02/15/2023 Our work on meta DAG structure learning for causal inference was accepted by ICASSP.
01/20/2023 1 paper was accepted by AISTATS and 2 papers were accepted by ICLR.
12/02/2022 Our work on conditional moment alignment for improved generalization in federated learning received the FL-NeurIPS Outstanding Paper Award!
11/28/2022 I am selected as the NeurIPS 2022 top reviewer.
10/21/2022 Two papers were accepted by NeurIPS Workshops.
11/17/2022 Two proposals were accepted by the Rensselaer-IBM Artificial Intelligence Research Collaboration program.
09/14/2022 Our paper on stochastic linearized augmented Lagrangian method for decentralized bilevel optimization was accepted by NeurIPS 2022.
09/14/2022 Our paper on understanding benign overfitting in nested meta learning was accepted by NeurIPS 2022.
09/07/2022 Our proposal titled "Adaptive, collaborative, and accelerated optimization for machine learning" was accepted by the MIT-IBM Watson AI Lab.
07/29/2022 Our work on distributed adversarial training to robustify deep neural networks at scale received the UAI 2022 Best Paper Runner-Up Award!
07/15/2022 Our work on achieving low sample and communication complexities in decentralized bilevel learning over networks was accepted by ACM MobiHoc 2022.
07/12/2022 Our work on bilevel optimization enhanced graph-aided federated learning was accepted by IEEE Transactions on Big Data.
06/14/2022 Our work on overcoming catastrophic forgetting via direction-constrained optimization was accepted by ECML PKDD 2022.
05/15/2022 Our work on distributed adversarial training to robustify deep neural networks at scale was accepted as an oral presentation at UAI 2022.
05/14/2022 My work on a single-loop gradient descent and perturbed ascent algorithm for nonconvex functional constrained optimization was accepted by ICML 2022.
04/20/2022 Our work on learning to generate image source-agnostic universal adversarial perturbations was accepted by IJCAI 2022.
02/28/2022 Our work on federated XGBoost using secret sharing and distributed optimization was accepted by ACM Transactions on Intelligent Systems and Technology.
02/27/2022 Our work on zeroth-order optimization for composite problems with functional constraints was selected as an oral presentation at AAAI 2022.
01/28/2022 Our work on finite-time convergence and sample complexity of multi-agent actor-critic reinforcement learning with average reward was selected as a spotlight (top 5%) at ICLR 2022.
01/28/2022 Our work on understanding latent correlation-based multiview learning and self-supervision: An identifiability perspective was selected as a spotlight (top 5%) at ICLR 2022.
01/21/2022 One paper was accepted by ICASSP 2022.
01/21/2022 Two papers were accepted by ICLR 2022.
12/07/2021 My work on "science of accurate, robust, and generalizable AI" received an IBM Research Accomplishment Award.
12/01/2021 Our paper adversarial examples for unsupervised machine learning models was accepted by AAAI 2022.
12/01/2021 Our paper on zeroth-order optimization for composite problems with functional constraints was accepted by AAAI 2022.
10/27/2021 Our paper on optimal discrete constellation inputs for aggregated LiFi-WiFi networks was accepted by IEEE Transactions on Wireless Communications.
09/29/2021 Our paper on sample- and communication-efficient policy evaluation for multi-agent reinforcement learning was accepted by NeurIPS 2021.
07/29/2021 I am honored to be the key personnel of the NSF AI Institute for future edge networks and distributed intelligence.
07/21/2021 I am honored to serve as a Senior Program Committee (SPC) Member for AAAI-22.
05/25/2021 Our paper entitled "Linearized ADMM converges to second-order stationary points for non-convex problems" wasw accepted by IEEE Transactions on Signal Processing (TSP).
03/16/2021 I am honored to serve as an FL-IJCAI'21 TPC member.
02/04/2021 Our paper entitled "Robust beamforming design for covert communications" was accepted by IEEE Transactions on Information Forensics and Security.
01/30/2021 Our paper entitled "Training logical neural networks by primal-dual methods for neuro-symbolic reasoning" was accepted by ICASSP 2021.
01/30/2021 Our paper entitled "Federated acoustic modeling for automatic speech recognition" was accepted by ICASSP 2021.
01/30/2021 Our paper entitled "On the convergence of randomized Bregman coordinate descent for non-Lipschitz composite problems" was accepted by ICASSP 2021.
01/22/2020 Our paper entitled "Rate-improved inexact augmented Lagrangian method for constrained nonconvex optimization" was accepted by the 24th International Conference on Artificial Intelligence and Statistics (AISTATS 2021).
12/02/2020 Our paper entitled "Decentralized policy gradient descent ascent for safe multi-agent reinforcement learning" was accepted by the Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI-21).
09/25/2020 Our paper entitled "Finding second-order stationary points efficiently in smooth nonconvex linearly constrained optimization problems" was accepted by the Thirty-fourth Conference on Neural Information Processing Systems (NeurIPS) as a Spotlight presentation, 2020.
09/25/2020 Our paper entitled "Decentralized TD tracking with linear function approximation and its finite-time analysis" was accepted by the Thirty-fourth Conference on Neural Information Processing Systems (NeurIPS), 2020.
09/25/2020 Our paper entitled "ScaleCom: Scalable sparsified gradient compression for communication-efficient distributed training" was accepted by the Thirty-fourth Conference on Neural Information Processing Systems (NeurIPS), 2020.
09/10/2020 I have been promoted as a (regular) Research Scientist in IBM Research AI.
08/18/2020 I will serve as a senior program committee (area chair) member for AAAI 2021.
06/15/2020 Our paper entitled "Non-convex min-max optimization: applications, challenges, and recent theoretical advances" was accepted by IEEE Signal Processing Magazine.
05/31/2020 Our paper entitled "Improving the sample and communication complexity for decentralized non-convex optimization: joint gradient estimation and tracking" was accepted by ICML 2020.
05/31/2020 Our paper entitled "Min-Max optimization without gradients: convergence and applications to black-box evasion and poisoning attacks" was accepted by ICML 2020.
03/28/2020 Our paper entitled "Achieving channel capacity of visible light communication" was accepted by IEEE Systems Journal.
01/24/2020 Two papers were accepted by ICASSP 2020.
01/05/2020 I was invited to give a talk about decentralized optimization at CISS.
01/05/2020 Our paper entitled "Distributed learning in the non-convex world: from batch to streaming data, and beyond " was accepted by IEEE Signal Processing Magazine.
12/18/2019 Our paper entitled "Hybrid block successive approximation for one-sided non-convex min-max problems: algorithms and applications " was accepted by IEEE Transactions on Signal Processing.
06/08/2019 I attended ICML at long beach from 6/8 to 6/14.
05/17/2019 Paper entitled "PA-GD: On the convergence of perturbed alternating gradient descent to second-order stationary points for structured nonconvex optimization" was selected as a Long Talk by ICML 2019. (20% of the accepted papers and 4.6% of the total submissions)
05/09/2019 I received the ICML 2019 travel award.
04/21/2019 Paper entitled "PA-GD: On the convergence of perturbed alternating gradient descent to second-order stationary points for structured nonconvex optimization" was accepted by ICML 2019.
04/12/2019 Paper entitled ''GNSD: A gradient-tracking based nonconvex stochastic algorithm for decentralized optimization'' was accepted by IEEE DSW 2019.
03/31/2019 I was invited to give a talk at INFORMS Annual Meeting 2019 (session: Recent Advances in Large-Scale Optimization).
02/07/2019 I was invited to give a talk at INFORMS Annual Meeting 2019 (session: Primal-Dual First-Order Methods).
02/01/2019 Three papers were accepted by IEEE ICASSP 2019.
01/21/2019 I was invited to give a presentation on the ITA 2019 Graduation Day.