I am a Research Scientist in AI/ML in Mobile Systems at Nokia Bell Labs Cambridge and a former PhD student at the Mobile Systems Research Lab, University of Cambridge. My work focuses on developing data-efficient and scalable machine learning algorithms, particularly for mobile systems. I specialize in developing robust AI models that can handle complex, real-world scenarios by leveraging data-efficient approaches, including semi-supervised and self-supervised learning.
My key research areas include:
- Scalable and Data-efficient Machine Learning for Human Activity Recognition: I develop novel training algorithms that reduce data dependence while ensuring robust recognition in mobile applications. My work leverages semi-supervised and self-supervised learning techniques for human activity recognition, utilizing methods such as contrastive learning [HCRL @ AAAI 2024, ML4MH @ NeurIPS 2020], self-training [IWMUT 2021], and multi-device collaboration [IWMUT 2022].
- Overcoming Catastrophic Forgetting in Continual Learning: I investigate strategies to address the challenges of catastrophic forgetting in continual learning, where models must learn from evolving data streams without losing previously acquired knowledge. My work focuses on approaches that balance stability and plasticity, ensuring models can generalize effectively across new tasks while retaining performance on past tasks [WACV 2024, ICASSP 2022].
- Federated Learning for Scalable Learning Algorithms: I explore decentralized machine learning paradigms that prioritize data privacy and enable collaborative learning across distributed devices [ICML 2022].
- AI for Health-related Applications: I apply machine learning techniques to health-related challenges, leveraging AI to enhance the accuracy and scalability of health monitoring applications [ML4H 2021, Nat. Mach. Intell. 2020].
I am passionate about leveraging AI to advance human-centric applications, from healthcare to mobile systems. By developing robust and scalable solutions, I aim to contribute to the future of ubiquitous computing, creating technologies that seamlessly integrate into and enhance our daily lives.
News
September 2024 - I will be running the second version of the UbiComp SOAR Tutorial on solving the activity recognition problem from 1:00 PM to 5:00 PM on October 6 at Melbourne Australia. Come and join us for an exciting discussion!
June 2022 - I am starting a research internship at Nokia Bell Labs (Cambridge), conducting research in machine learning on heterogenous edge devices.
May 2022 - Source code for "Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering" has been released on GitHub.
May 2022 - Paper "Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering" has been accepted for presentation in ICML 2022 and is now available on arXiv.
May 2022 - Presented "Improving Feature Generalizability with Multitask Learning in Class Incremental Learning" at ICASSP 2022. Recorded talk available here.
March 2022 - Paper "ColloSSL: Collaborative Self-Supervised Learning for Human Activity Recognition" has been published in the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT).
March 2022 - A set of notes on Lambda calculus "How to Use the Y Combinator" has been uploaded to this website, as complementary materials for the course Computation Theory (2021-2022).
June 2021 - Source code for "SelfHAR: Improving Human Activity Recognition through Self-training with Unlabeled Data" has been released on GitHub.
April 2021 - Paper "SelfHAR: Improving Human Activity Recognition through Self-training with Unlabeled Data" has been published in the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT).
Works/Publications
2024
Solving the Sensor-Based Activity Recognition Problem (SOAR): Self-Supervised, Multi-Modal Recognition of Activities from Wearable Sensors
Harish Haresamudram, Chi Ian Tang, Sungho Suh, Paul Lukowicz, Thomas Ploetz
In UbiComp/ISWC 2024 Adjunct
Balancing Continual Learning and Fine-tuning for Human Activity Recognition
Chi Ian Tang, Lorena Qendro, Dimitris Spathis, Fahim Kawsar, Akhil Mathur, Cecilia Mascolo
In AAAI 2024 Workshop: Human-Centric Representation Learning (HCRL)
Kaizen: Practical Self-Supervised Continual Learning With Continual Fine-Tuning
Chi Ian Tang, Lorena Qendro, Dimitris Spathis, Fahim Kawsar, Cecilia Mascolo, Akhil Mathur
In WACV 2024 (IEEE/CVF Winter Conference on Applications of Computer Vision)
2023
Self-supervised Learning for Data-efficient Human Activity Recognition
Chi Ian Tang
PhD Thesis
Solving the Sensor-based Activity Recognition Problem (SOAR): Self-supervised, Multi-modal Recognition of Activities from Wearable Sensors
Harish Haresamudram, Chi Ian Tang, Sungho Suh, Paul Lukowicz, Thomas Ploetz
In UbiComp/ISWC 2023 Adjunct
2022
Orchestra: Unsupervised Federated Learning via Globally Consistent Clustering
Ekdeep Singh Lubana, Chi Ian Tang, Fahim Kawsar, Robert P. Dick, Akhil Mathur
In ICML 2022 (International Conference on Machine Learning)
Improving Feature Generalizability with Multitask Learning in Class Incremental Learning
Dong Ma*, Chi Ian Tang*, Cecilia Mascolo
*Ordered alphabetically, equal contribution
In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022
ColloSSL: Collaborative Self-Supervised Learning for Human Activity Recognition
Yash Jain*, Chi Ian Tang*, Chulhong Min, Fahim Kawsar, Akhil Mathur
*Ordered alphabetically, equal contribution
In Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT). Volume 6 Issue 1, Article 17 (March 2022).
2021
Evaluating Contrastive Learning on Wearable Timeseries for Downstream Clinical Outcomes
Kevalee Shah, Dimitris Spathis, Chi Ian Tang, Cecilia Mascolo
In Machine Learning for Health (ML4H) 2021
Group Supervised Learning: Extending Self-Supervised Learning to Multi-Device Settings
Yash Jain*, Chi Ian Tang*, Chulhong Min, Fahim Kawsar, Akhil Mathur
*Equal Contribution
In ICML 2021 Workshop: Self-Supervised Learning for Reasoning and Perception
SelfHAR: Improving Human Activity Recognition through Self-training with Unlabeled Data
Chi Ian Tang, Ignacio Perez-Pozuelo, Dimitris Spathis, Soren Brage, Nick Wareham, Cecilia Mascolo.
In Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT). Volume 5 Issue 1, Article 36 (March 2021).
2020
Exploring Contrastive Learning in Human Activity Recognition for Healthcare
Chi Ian Tang, Dimitris Spathis, Ignacio Perez Pozuelo, Cecilia Mascolo.
In ML for Mobile Health Workshop at NeurIPS. 2020.
Exploring the limit of using a deep neural network on pileup data for germline variant calling
Ruibang Luo, Chak-Lim Wong, Yat-Sing Wong, Chi-Ian Tang, Chi-Man Liu, Henry CM Leung, Tak-Wah Lam.
In Nature Machine Intelligence. 2020.
Academic Service
I have taking up organizing roles for the following:
- Organizer of UbiComp Tutorial on Solving the sensor-based activity recognition problem (SOAR) (2023 Cancun, Mexico, 2024 Melbourne, Australia)
- Organizer of HCRL workshop at AAAI 2024, Vancouver, Canada
Teaching
Lectures/Tutorials
I gave a lecture/tutorial on Machine Learning and Features of Health Data in the Mobile Health course (Master's level) at the Department of Computer Science and Technology, University of Cambridge. Slides are available Slides (2023)
Supervisions
I have supervised students and demonstrated for the following courses at the University of Cambridge:
- Programming in C and C++ (2020-2021)
- Machine Learning and Real-world Data (2019-2021)
- Computation Theory (2021-2024)
- Digital Electronics (2022-2023)
Teaching Materials
- How to Use the Y Combinator (for Computation Theory 2021-2022)
© 2020-2024 Chi Ian Tang. All rights reserved.