Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Transfer Hashing for Adaptive Learning on Federated Streaming Data

Manuel Röder, Frank-Michael Schleif

TL;DR

The paper addresses the challenge of privacy-preserving, efficient learning on federated streaming data by combining federated learning with deep transfer hashing and transfer learning. It proposes server-side pre-training of a hash function $h$ and a privacy-preserving global memory bank $M_S$ to support on-device fine-tuning, reducing data transmission and enabling rapid adaptation to evolving streams. A SHAR-inspired local adaptation pattern and selective hash-code sharing are introduced to improve communication efficiency and scalability in real-world scenarios such as Car2X deployments, where traffic-related tasks require robust, online adaptation. The approach aims to deliver practical, secure downstream classification and retrieval with improved efficiency, addressing concept drift and privacy constraints in distributed settings.

Abstract

This extended abstract explores the integration of federated learning with deep transfer hashing for distributed prediction tasks, emphasizing resource-efficient client training from evolving data streams. Federated learning allows multiple clients to collaboratively train a shared model while maintaining data privacy - by incorporating deep transfer hashing, high-dimensional data can be converted into compact hash codes, reducing data transmission size and network loads. The proposed framework utilizes transfer learning, pre-training deep neural networks on a central server, and fine-tuning on clients to enhance model accuracy and adaptability. A selective hash code sharing mechanism using a privacy-preserving global memory bank further supports client fine-tuning. This approach addresses challenges in previous research by improving computational efficiency and scalability. Practical applications include Car2X event predictions, where a shared model is collectively trained to recognize traffic patterns, aiding in tasks such as traffic density assessment and accident detection. The research aims to develop a robust framework that combines federated learning, deep transfer hashing and transfer learning for efficient and secure downstream task execution.

Deep Transfer Hashing for Adaptive Learning on Federated Streaming Data

TL;DR

The paper addresses the challenge of privacy-preserving, efficient learning on federated streaming data by combining federated learning with deep transfer hashing and transfer learning. It proposes server-side pre-training of a hash function and a privacy-preserving global memory bank to support on-device fine-tuning, reducing data transmission and enabling rapid adaptation to evolving streams. A SHAR-inspired local adaptation pattern and selective hash-code sharing are introduced to improve communication efficiency and scalability in real-world scenarios such as Car2X deployments, where traffic-related tasks require robust, online adaptation. The approach aims to deliver practical, secure downstream classification and retrieval with improved efficiency, addressing concept drift and privacy constraints in distributed settings.

Abstract

This extended abstract explores the integration of federated learning with deep transfer hashing for distributed prediction tasks, emphasizing resource-efficient client training from evolving data streams. Federated learning allows multiple clients to collaboratively train a shared model while maintaining data privacy - by incorporating deep transfer hashing, high-dimensional data can be converted into compact hash codes, reducing data transmission size and network loads. The proposed framework utilizes transfer learning, pre-training deep neural networks on a central server, and fine-tuning on clients to enhance model accuracy and adaptability. A selective hash code sharing mechanism using a privacy-preserving global memory bank further supports client fine-tuning. This approach addresses challenges in previous research by improving computational efficiency and scalability. Practical applications include Car2X event predictions, where a shared model is collectively trained to recognize traffic patterns, aiding in tasks such as traffic density assessment and accident detection. The research aims to develop a robust framework that combines federated learning, deep transfer hashing and transfer learning for efficient and secure downstream task execution.
Paper Structure (3 sections, 1 figure, 1 algorithm)

This paper contains 3 sections, 1 figure, 1 algorithm.

Table of Contents

  1. Introduction and Background
  2. Methodology
  3. Conclusion and Open Questions

Figures (1)

  • Figure 1: Car2x scenario: FL clients with monitoring sensors (camera, radar) sample from a traffic input stream at different geographic locations. The central server instance maintains a global memory bank that is accessible for all client devices to support local client model fine-tuning. The feature extractor and hashing layer are pre-trained on the large-scale server data set $D_S$. The central server administrates a global memory bank.