A privacy-preserving data valuation visualization system

• Main Author: Yap, Rong Yu
• Other Authors: Yu Han
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/153309
• Institution: Nanyang Technological University

Data is increasingly being regulated by the governments, making it difficult to conduct collaborative machine learning without violating the regulation. This leads to the increased interest in federated learning as data is processed at the client-side. However, stakeholders are hesitant to participate in federated learning. This is due to federated learning producing a huge amount of data as output and thus it is difficult to interpret the results of federated learning. This leads to a need to have a visualisation system to present data in a manner that the stakeholders can interpret. Current visualisation systems are unable to meet the needs of the stakeholders as they are not able to handle the large data output produced by federated learning. In this report, Shapley value and its various estimation will be reviewed along with the previous studies of visualisation systems for federated learning. The design and results of the visualisation system will be discussed after the literature review.