| Author: | Liu, Peng |
| Title: | Virtual digital human and its applications in finance industry |
| Advisors: | You, Jane (ISE) |
| Degree: | Eng.D. |
| Year: | 2026 |
| Department: | Department of Computing |
| Pages: | xv, 170 pages : color illustrations |
| Language: | English |
| Abstract: | Customers are a key treasure to the enterprise. Enterprise tends to provide better services to customers in order to not only optimize customer satisfaction but also increase potential sales. Interactions between customers and the enterprise are crucial in achieving these goals. The interactions can be categorized into two types: inbound and outbound. Artificial intelligence (AI) has been a significant development, and tremendous algorithms have been proposed by academia and applied in industry after mature verification. These algorithms are related to the standard modalities—such as text, audio, and video—that characterize our daily human interactions. More specifically, Natural Language Processing (NLP) can build tools to understand and generate human language text, Acoustic Speech Recognition (ASR) can convert utterance audio into text, and Text-to-Speech (TTS) can generate human-like speech from text, generative models such as Generative Adversarial Networks (GANs) and Diffusion Models can create high fidelity visual content such as images and videos. This thesis will present a comprehensive study linking the trending development of AI and the demand for interactions. Ping An Insurance Group and its subsidiaries are selected as representative samples of the finance industry for the study since the author belongs to its core innovation technology subsidiary. Historically, the subsidiaries had two evolving generation eras to enhance the interactions at the beginning of this study. The first era emphasized text-based chatbots, which handled inbound interactions only. The second era emphasized audio-based phone call interactions, including inbound and outbound interactions. The third era emerges within this study and emphasizes video-based interactions in conjunction with the flourishing self-media platforms such as WeChat Video Account, Douyin, and Little Red Book. The core concept of this interaction type is the virtual digital human, which acts as an upgraded version of the chatbot and Interactive Voice Representative (IVR). The author includes an under-reviewed virtual digital human research paper in this thesis. The author also includes a published Neural Network Search (NAS) conference paper as an auxiliary approach to virtual digital human algorithm training. Due to the strict law and regulation constraints of handling sensitive data (authentic human images and videos of different subsidiaries), the author includes a published conference paper on Federated Learning. Not only the academic research aspect but also the engineering implementation and design aspects are included in this thesis. Finally, some desensitized achievement information is disclosed to demonstrate the commercial value related to this study. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 9082.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 18.57 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/14594