HeteroSwitch Dataset
HeteroSwitch Dataset
Datasets
HeteroSwitch Dataset
File
HeteroSwitch Dataset
Use Case
Computer Vision
Description
Explore the HeteroSwitch Dataset, designed for research on system-induced data heterogeneity in Federated Learning environments.
Research Context
The HeteroSwitch research, detailed in “HeteroSwitch: Characterizing and Taming System-Induced Data Heterogeneity in Federated Learning,” explores the challenges posed by data heterogeneity in Federated Learning (FL) systems. This research has been published on arXiv and presented at the 7th Annual Conference on Machine Learning and Systems (MLSys 2024), providing key insights into managing data inconsistencies across diverse systems.
Dataset Description
It includes a wide array of data collected from multiple types of devices, focusing on the effects of system variability on data uniformity.
Data Collection Methodology
To assess the influence of system-induced heterogeneity on FL, the dataset was generated using nine different smartphone models: Galaxy S22, Galaxy S9, Galaxy S6, VELVET, G7, G4, Pixel 5, Pixel 2, and Nexus 5X. Each smartphone was mounted on a tripod and used to capture images displayed on a monitor within a controlled darkroom setting.
Dataset Contents
The dataset consists of images from 12 distinct ImageNet classes, each representing different higher-level categories. The selected classes include Chihuahua, Altar, Cock, Abaya, Ambulance, Loggerhead, Timber Wolf, Tiger Beetle, Accordion, French Loaf, Barber Chair, and Orangutan. These classes were chosen to avoid overlap and to represent a broad spectrum of visual features.
Contact Us
Quality Data Creation
Guaranteed TAT
ISO 9001:2015, ISO/IEC 27001:2013 Certified
HIPAA Compliance
GDPR Compliance
Compliance and Security
Let's Discuss your Data collection Requirement With Us
To get a detailed estimation of requirements please reach us.