Distributed Machine Learning Algorithms for Resource-Constrained Heterogeneous Internet-of-Things Environments

Speaker Bio

Ahmed Imteaj is currently a Ph.D. Candidate at the Knight Foundation School of Computing and Information Sciences at Florida International University, working with Prof. M. Hadi Amini. His research interests span Federated Learning, Internet of Things (IoT), Artificial Intelligence, and Blockchain, and he has several peer-reviewed journals and conference publications in top venues. Ahmed received his B.Sc. degree in Computer Science and Engineering from CUET, Bangladesh, and received his Master’s in Computer Science from FlU with the recognition of the “Outstanding Master’s degree graduate Award”. He also received the “2022 Outstanding Student Life Award” (the Graduate Scholar of the Year Award) from the Division of Academic and Student Affairs at FIU, “2021 Best Graduate Student in Research Award” from the Knight Foundation School of Computing and Information Sciences at FIU, is the recipient of the Best Paper Award from the 2019 IEEE CSCI’19 conference, recipient of 3 travel scholarships, and also one of the two winners at 2021 Florida International University Graduate Student Appreciation Week. During his PhD, Ahmed mentored 8 undergraduate and graduate students, and was actively involved in NSF REU and RET programs. Ahmed is also the lead author of the book, “Foundations of Blockchain: Theory and Applications”.

Description

With the improvement of network infrastructures and the advancement of IoT technologies, it is crucial to perform computation at the edges, rather than sharing data with fusion centers, which is privacy-intrusive. Centralized machine learning (ML) algorithms fail to address underlying challenges related to users’ privacy or capture global knowledge of the whole network. To properly handle such challenges, a recently invented distributed ML technique, called Federated Learning (FL) was invented as a means to construct a global model without exposing users’ private data through on-device model training utilizing edge resources. However, FL may face various challenges due to the lack of a convenient mechanism to prepare a federated dataset, and also the heterogeneous nature of the resource-constrained agents such as systems, statistical and model heterogeneity. We developed a distributed sensing mechanism through which any federated agents can be triggered and activated for sensing the environment. That novel approach shows a pathway to carry out the FL process in a real-world environment. Following this, we developed an FL model, FedAR by monitoring agent activities and leveraging available local computing resources, particularly for resource-constrained IoT devices (e.g., mobile robots), to accelerate the learning process. After that, we propose a tri-layer FL framework, FedPARL that helps resource-constrained FL agents consume less resources during training, avoid untrustworthy and out-of-resource agents (e.g., low battery life) during agent selection for training, and perform variable local epochs based on the agent’s resource availability. We perform model pruning to reduce the size of the agent model that is effective for an FL-IoT setting. In order to evaluate our proposed distributed ML algorithms, we have explored three real-world applications: i) using FL to improve the resilience of critical infrastructures through knowledge exchange; ii) Using FL to recognize human activities; and iii) forecast customers’ financial distress in resource-constrained environments. These applications verify the performance of the FL-based algorithms in real-world settings considering resource limitations.

View on Zoom: https://fiu.zoom.us/j/99501006064?pwd=dG1uOVZvMkpYdytBeW5RckVDSkljdz09
Meeting ID: 995 0100 6064
Passcode: u5nCda