University of Twente Student Theses

Login
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.

P-FedNIP: A Multi-Layered Personalized Federated Learning Framework

Nanduri, Rahul Srivatsa Sai (2025) P-FedNIP: A Multi-Layered Personalized Federated Learning Framework.

[img] PDF
3MB
Abstract:Energy forecasting presents unique challenges due to temporal dependencies, seasonal variations, and di- verse consumption patterns that require specialized fed- erated learning approaches to capture local patterns while preserving data privacy. While Federated Learn- ing (FL) has emerged as a privacy-preserving alterna- tive for training models on decentralized data, standard algorithms like FedAvg often falter under the non-IID conditions inherent in energy consumption data, lead- ing to reduced convergence speed and suboptimal model accuracy. To address these limitations, we propose P- FedNIP, a novel multi-layered personalized federated learning framework. P-FedNIP extends the FedNIP algorithm by introducing a sophisticated architecture that combines (1) EMD-based client clustering to un- derstand the data landscape, (2) intelligent client selec- tion to optimize training, (3) FedProx regularization to prevent local model drift, and (4) adaptive fine-tuning for deep personalization. This approach aims to create both a robust global model and highly accurate person- alized models tailored to the unique energy consump- tion patterns of each participant.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business & IT BSc (56066)
Link to this item:https://purl.utwente.nl/essays/107873
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page