Publications
Energy and Latency-aware Computation Load Distribution of Hybrid Split and Federated Learning on IoT Devices - Details
In Proceedings of the 10th International Conference on Networking, Systems and Security (NSysS ’23), 2023
In this paper, we develop an adaptive clustering-based computation load distribution method for client devices in hybrid Split and Federated Learning on IoT devices, with heterogeneous resource capacities, participating in the model training.
Recommended citation:
Sakhaouth Hossan, Farhan Mahmud, Palash Roy, Md Abdur Razzaque, and Md Mustafizur Rahman. Energy and Latency-aware Computation Load Distribution of Hybrid Split and Federated Learning on IoT devices. In Proceedings of the 10th International Conference on Networking, Systems and Security, pages 61–68, 2023.
https://doi.org/10.1145/3629188.3629201
An Insight into Dhaka City’s Walkability
Publicly available dataset on Kaggle, 2023
Crowdsourced research project of a team of ~70 to collect and analyze a crowdsourced dataset on the walkability of Dhaka City. The dataset, available on Kaggle, provides insights into factors influencing walking experiences in the city.
Ongoing Research Projects
LULC Analysis of Bangladesh using Deep Learning - Details
Land use land cover (LULC) is defined how the land is used by humans or what the surface is covered with. The goal of this project is to produce good quality LULC annotation data pixel by pixel for Dhaka city and surrounding area. With a sufficiently trained segmentation model, we can employ this model to perform LULC over the entire Bangladesh without any human effort.
Supervisor: AKM Mahbubur Rahman
Resource Aware Anonymous Clustered/Tiered Federated Learning - Details
Resource-aware clustering for federated learning is usually done by the server collecting device computation and communication resource information, which is a sensitive info. In this, work we are developing a clustered FL framework which doesn’t expose device information to server.
Supervisor: Sajib Mistry
Federated Learning For Satellite Telemetry - Details
This work is on developing an efficient federated learning framework in resource constrained satellite environments for anomaly detection from satellite telemetry data.
Supervisor: Sajib Mistry