Lili Su
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Lili Su has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
- Carlee Joe-Wong
Carnegie Mellon University, Pittsburgh, USA
8 shared publications - Edmund Yeh
Department of Electrical and Computer Engineering, Northeastern University, Boston, MA, USA
7 shared publications - Stratis Ioannidis
Northeastern University, Boston, MA, USA
7 shared publications - Aidong Adam Ding
Northeastern University, Boston, USA
3 shared publications - Jiaming Xu
The Fuqua School of Business, Duke University, Durham, North Carolina 27708
3 shared publications - Marie Siew
Information Systems and Technology and Design Pillar, Singapore University of Technology and Design, Singapore
3 shared publications
Latest Publications
- Open Access
- 22 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 61-70
Over-the-Air Federated Learning in Heterogeneous Mobile Wireless Networks
- 22 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 61-70
Open AccessOver-the-Air Federated Learning in Heterogeneous Mobile Wireless Networks
- Open Access
- 26 March 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Cyber-Physical Systems
Building Real-time Awareness of Out-of-distribution in Trajectory Prediction for Autonomous Vehicles
Embodied AI systems operate in a closed perception-action loop. In many safety-critical applications, embodied AI systems must adapt to highly uncertain and dynamically evolving environments. In the context of autonomous vehicles, trustworthy trajectory prediction is therefore a prerequisite for safe interaction with surrounding agents and infrastructure. In practice, however, even well-trained machine learning models may produce unreliable predictions. Such sim-to-real gaps (also known as imperfect training data) may be unavoidable due to the overwhelming complexity of data annotation and environment uncertainties. To ensure safety, it is critical to quickly detect when a model's predictions become unreliable. Leveraging the intuition that in-distribution (ID) scenes exhibit error patterns similar to training data, while out-of-distribution (OOD) scenes do not, we introduce a principled and computationally efficient approach for OOD detection by framing it as a quickest change-point detection problem. We address the challenging settings where the OOD scenes are deceptive, meaning that they are not easily detectable by human intuitions. Our solutions can handle the occurrence of OOD at any time during trajectory prediction inference. Experimental results on multiple real-world datasets demonstrate the effectiveness of our methods.
- 26 March 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Cyber-Physical Systems
Open AccessBuilding Real-time Awareness of Out-of-distribution in Trajectory Prediction for Autonomous Vehicles
Embodied AI systems operate in a closed perception-action loop. In many safety-critical applications, embodied AI systems must adapt to highly uncertain and dynamically evolving environments. In the context of autonomous vehicles, trustworthy trajectory prediction is therefore a prerequisite for safe interaction with surrounding agents and infrastructure. In practice, however, even well-trained machine learning models may produce unreliable predictions. Such sim-to-real gaps (also known as imperfect training data) may be unavoidable due to the overwhelming complexity of data annotation and environment uncertainties. To ensure safety, it is critical to quickly detect when a model's predictions become unreliable. Leveraging the intuition that in-distribution (ID) scenes exhibit error patterns similar to training data, while out-of-distribution (OOD) scenes do not, we introduce a principled and computationally efficient approach for OOD detection by framing it as a quickest change-point detection problem. We address the challenging settings where the OOD scenes are deceptive, meaning that they are not easily detectable by human intuitions. Our solutions can handle the occurrence of OOD at any time during trajectory prediction inference. Experimental results on multiple real-world datasets demonstrate the effectiveness of our methods.
- 1 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automatic Control
- Vol. 71  (3) , 1692-1707
Fast and Robust State Estimation and Tracking via Hierarchical Learning
- 1 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automatic Control
- Vol. 71  (3) , 1692-1707
Fast and Robust State Estimation and Tracking via Hierarchical Learning
- 23 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 33  (6) , 2881-2896
Fair Concurrent Training of Multiple Models in Federated Learning
Federated learning (FL) enables collaborative learning across multiple clients. In most FL work, all clients train a single learning task. However, the recent proliferation of FL applications may increasingly require multiple FL tasks to be trained simultaneously, sharing clients’ computing resources, which we call Multiple-Model Federated Learning (MMFL). Current MMFL algorithms use naïve average-based client-task allocation schemes that often lead to unfair performance when FL tasks have heterogeneous difficulty levels, as the more difficult tasks may need more client participation to train effectively. Furthermore, in the MMFL setting, we face a further challenge that some clients may prefer training specific tasks to others, and may not even be willing to train other tasks, e.g., due to high computational costs, which may exacerbate unfairness in training outcomes across tasks. We address both challenges by firstly designing FedFairMMFL, a difficulty-aware algorithm that dynamically allocates clients to tasks in each training round, based on the tasks’ current performance levels. We provide guarantees on the resulting task fairness and FedFairMMFL’s convergence rate. We then propose novel auction designs that incentivizes clients to train multiple tasks, so as to fairly distribute clients’ training efforts across the tasks, and extend our convergence guarantees to this setting. We finally evaluate our algorithm with multiple sets of learning tasks on real world datasets, showing that our algorithm improves fairness by improving the final model accuracy and convergence speed of the worst performing tasks, while maintaining the average accuracy across tasks.
- 23 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 33  (6) , 2881-2896
Fair Concurrent Training of Multiple Models in Federated Learning
Federated learning (FL) enables collaborative learning across multiple clients. In most FL work, all clients train a single learning task. However, the recent proliferation of FL applications may increasingly require multiple FL tasks to be trained simultaneously, sharing clients’ computing resources, which we call Multiple-Model Federated Learning (MMFL). Current MMFL algorithms use naïve average-based client-task allocation schemes that often lead to unfair performance when FL tasks have heterogeneous difficulty levels, as the more difficult tasks may need more client participation to train effectively. Furthermore, in the MMFL setting, we face a further challenge that some clients may prefer training specific tasks to others, and may not even be willing to train other tasks, e.g., due to high computational costs, which may exacerbate unfairness in training outcomes across tasks. We address both challenges by firstly designing FedFairMMFL, a difficulty-aware algorithm that dynamically allocates clients to tasks in each training round, based on the tasks’ current performance levels. We provide guarantees on the resulting task fairness and FedFairMMFL’s convergence rate. We then propose novel auction designs that incentivizes clients to train multiple tasks, so as to fairly distribute clients’ training efforts across the tasks, and extend our convergence guarantees to this setting. We finally evaluate our algorithm with multiple sets of learning tasks on real world datasets, showing that our algorithm improves fairness by improving the final model accuracy and convergence speed of the worst performing tasks, while maintaining the average accuracy across tasks.
- Open Access
- 30 May 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Modeling and Performance Evaluation of Computing Systems
- Vol. 10  (2) , 1-2
Introduction to the Special Issue on Performance Evaluation of Federated Learning Systems Part 2
- 30 May 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Modeling and Performance Evaluation of Computing Systems
- Vol. 10  (2) , 1-2
Open AccessIntroduction to the Special Issue on Performance Evaluation of Federated Learning Systems Part 2
- Open Access
- 12 March 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Modeling and Performance Evaluation of Computing Systems
- Vol. 10  (1) , 1-2
Introduction to the Special Issue on Performance Evaluation of Federated Learning Systems Part 1
- 12 March 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Modeling and Performance Evaluation of Computing Systems
- Vol. 10  (1) , 1-2
Open AccessIntroduction to the Special Issue on Performance Evaluation of Federated Learning Systems Part 1
- 16 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 73  (1053587X) , 766-780
Empowering Federated Learning With Implicit Gossiping: Mitigating Connection Unreliability Amidst Unknown and Arbitrary Dynamics
Federated learning is a popular distributed learning approach for training a machine learning model without disclosing raw data. It consists of a parameter server and a possibly large collection of clients (e.g., in cross-device federated learning) that may operate in congested and changing environments. In this paper, we study federated learning in the presence of stochastic and dynamic communication failures wherein the uplink between the parameter server and client i is on with unknown probability pti in round t. Furthermore, we allow the dynamics of pti to be arbitrary. We first demonstrate that when the pti’s vary across clients, the most widely adopted federated learning algorithm, Federated Average (FedAvg), experiences significant bias. To address this observation, we propose Federated Postponed Broadcast (FedPBC), a simple variant of FedAvg. It differs from FedAvg in that the parameter server postpones broadcasting the global model to the clients with active uplinks till the end of each training round. Despite uplink failures, we show that FedPBC converges to a stationary point of the original non-convex objective. On the technical front, postponing the global model broadcasts enables implicit gossiping among the clients with active links in round t. In spite of the time-varying nature of pti , we can bound the perturbation of the global model dynamics using techniques to control gossip-type information mixing errors. Extensive experiments have been conducted on real-world datasets over diversified unreliable uplink patterns to corroborate our analysis.
- 16 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Signal Processing
- Vol. 73  (1053587X) , 766-780
Empowering Federated Learning With Implicit Gossiping: Mitigating Connection Unreliability Amidst Unknown and Arbitrary Dynamics
Federated learning is a popular distributed learning approach for training a machine learning model without disclosing raw data. It consists of a parameter server and a possibly large collection of clients (e.g., in cross-device federated learning) that may operate in congested and changing environments. In this paper, we study federated learning in the presence of stochastic and dynamic communication failures wherein the uplink between the parameter server and client i is on with unknown probability pti in round t. Furthermore, we allow the dynamics of pti to be arbitrary. We first demonstrate that when the pti’s vary across clients, the most widely adopted federated learning algorithm, Federated Average (FedAvg), experiences significant bias. To address this observation, we propose Federated Postponed Broadcast (FedPBC), a simple variant of FedAvg. It differs from FedAvg in that the parameter server postpones broadcasting the global model to the clients with active uplinks till the end of each training round. Despite uplink failures, we show that FedPBC converges to a stationary point of the original non-convex objective. On the technical front, postponing the global model broadcasts enables implicit gossiping among the clients with active links in round t. In spite of the time-varying nature of pti , we can bound the perturbation of the global model dynamics using techniques to control gossip-type information mixing errors. Extensive experiments have been conducted on real-world datasets over diversified unreliable uplink patterns to corroborate our analysis.
- Open Access
Probe-Me-Not: Protecting Pre-trained Encoders from Malicious Probing
Adapting pre-trained deep learning models to customized tasks has become a popular choice for developers to cope with limited computational resources and data volume. More specifically, probing--training a downstream head on a pre-trained encoder--has been widely adopted in transfer learning, which helps to prevent overfitting and catastrophic forgetting. However, such generalizability of pre-trained encoders raises concerns about the potential misuse of probing for harmful intentions, such as discriminatory speculation and warfare applications. In this work, we introduce EncoderLock, a novel applicability authorization method designed to protect pre-trained encoders from malicious probing, i.e., yielding poor performance on specified prohibited domains while maintaining their utility in authorized ones. Achieving this balance is challenging because of the opposite optimization objectives and the variety of downstream heads that adversaries can utilize adaptively. To address these challenges, EncoderLock employs two techniques: domain-aware weight selection and updating to restrict applications on prohibited domains/tasks, and self-challenging training scheme that iteratively strengthens resistance against any potential downstream classifiers that adversaries may apply. Moreover, recognizing the potential lack of data from prohibited domains in practical scenarios, we introduce three EncoderLock variants with different levels of data accessibility: supervised (prohibited domain data with labels), unsupervised (prohibited domain data without labels), and zero-shot (no data or labels available). We verify EncoderLock's effectiveness and practicality with a real-world pre-trained Vision Transformer (ViT) encoder from Facebook. These results underscore the valuable contributions EncoderLock brings to the development of responsible AI.
Open AccessProbe-Me-Not: Protecting Pre-trained Encoders from Malicious Probing
Adapting pre-trained deep learning models to customized tasks has become a popular choice for developers to cope with limited computational resources and data volume. More specifically, probing--training a downstream head on a pre-trained encoder--has been widely adopted in transfer learning, which helps to prevent overfitting and catastrophic forgetting. However, such generalizability of pre-trained encoders raises concerns about the potential misuse of probing for harmful intentions, such as discriminatory speculation and warfare applications. In this work, we introduce EncoderLock, a novel applicability authorization method designed to protect pre-trained encoders from malicious probing, i.e., yielding poor performance on specified prohibited domains while maintaining their utility in authorized ones. Achieving this balance is challenging because of the opposite optimization objectives and the variety of downstream heads that adversaries can utilize adaptively. To address these challenges, EncoderLock employs two techniques: domain-aware weight selection and updating to restrict applications on prohibited domains/tasks, and self-challenging training scheme that iteratively strengthens resistance against any potential downstream classifiers that adversaries may apply. Moreover, recognizing the potential lack of data from prohibited domains in practical scenarios, we introduce three EncoderLock variants with different levels of data accessibility: supervised (prohibited domain data with labels), unsupervised (prohibited domain data without labels), and zero-shot (no data or labels available). We verify EncoderLock's effectiveness and practicality with a real-world pre-trained Vision Transformer (ViT) encoder from Facebook. These results underscore the valuable contributions EncoderLock brings to the development of responsible AI.
- Open Access
Personalized Federated Learning via Feature Distribution Adaptation
Federated learning (FL) is a distributed learning framework that leverages commonalities between distributed client datasets to train a global model. Under heterogeneous clients, however, FL can fail to produce stable training results. Personalized federated learning (PFL) seeks to address this by learning individual models tailored to each client. One approach is to decompose model training into shared representation learning and personalized classifier training. Nonetheless, previous works struggle to navigate the bias-variance trade-off in classifier learning, relying solely on limited local datasets or introducing costly techniques to improve generalization. In this work, we frame representation learning as a generative modeling task, where representations are trained with a classifier based on the global feature distribution. We then propose an algorithm, pFedFDA, that efficiently generates personalized models by adapting global generative classifiers to their local feature distributions. Through extensive computer vision benchmarks, we demonstrate that our method can adjust to complex distribution shifts with significant improvements over current state-of-the-art in data-scarce settings.
Open AccessPersonalized Federated Learning via Feature Distribution Adaptation
Federated learning (FL) is a distributed learning framework that leverages commonalities between distributed client datasets to train a global model. Under heterogeneous clients, however, FL can fail to produce stable training results. Personalized federated learning (PFL) seeks to address this by learning individual models tailored to each client. One approach is to decompose model training into shared representation learning and personalized classifier training. Nonetheless, previous works struggle to navigate the bias-variance trade-off in classifier learning, relying solely on limited local datasets or introducing costly techniques to improve generalization. In this work, we frame representation learning as a generative modeling task, where representations are trained with a classifier based on the global feature distribution. We then propose an algorithm, pFedFDA, that efficiently generates personalized models by adapting global generative classifiers to their local feature distributions. Through extensive computer vision benchmarks, we demonstrate that our method can adjust to complex distribution shifts with significant improvements over current state-of-the-art in data-scarce settings.