Abstract

The widespread deployment of wireless and mobile devices results in a proliferation of spatio- temporal data (e.g., time series, streaming, and trajectory data) that is essential for applications, e.g., traffic and air quality prediction, where spatio-temporal analytics plays a key role in ensuring safety, predictability, and reliability. While recent studies have demonstrated superior performance in deep spatio-temporal analytics, many approaches struggle to adapt to real-world conditions. In particular, existing methods suffer from three limitations: 1) existing deep methods typically require large-scale training data, incurring high storage and computational costs; 2) when applied to streaming data, many models suffer from catastrophic forgetting, resulting in deteriorating prediction accuracy over time; and 3) existing solutions often assume centralized data, which leads to privacy concerns and fails to exploit decentralized data processing.

This Ph.D. study aims to systematically study deep spatio-temporal analytics with emerging techniques. Specifically, we target four types of functionality: time series dataset condensation, streaming spatio-temporal prediction, federated trajectory recovery, and federated trajectory similarity learning.

First, we address the problem of time series dataset condensation. The goal is to reduce training costs by summarizing large datasets into smaller, synthetic datasets that can then be used for training instead. We introduce TimeDC, an efficient time series dataset condensation framework that uses two-fold modal matching. TimeDC encompasses a time series feature extraction module for effective feature learning, a decomposition-driven frequency matching module for achieving temporal correlation preservation, and a curriculum- based trajectory matching module for ensuring that the synthetic datasets capture key patterns in the original dataset.

Second, we investigate spatio-temporal prediction on streaming data. We propose URCL, a unified replay-based streaming framework with three key modules: data integration, spatio-temporal continuous representation learning, and spatio-temporal prediction. Specifically, a spatio-temporal mixup mechanism is introduced to alleviate catastrophic forgetting, and a simple siamese network is designed to facilitate holistic feature learning.

Third, we study the problem of federated trajectory recovery, focusing on privacy preservation and enabling decentralized training. We propose LighTR+, a horizontal federated framework, which consists of a lightweight local trajectory embedding module, an intra-domain knowledge distillation module, and a meta-knowledge enhanced local-global training scheme. LighTR+ alleviates intra- and inter-domain interferences across clients while reducing communication costs between clients and the server, thereby facilitating privacy protection and improving efficiency.

Fourth, we explore federated trajectory similarity learning for decentralized data processing. We propose the F-TSL framework based on horizontal federated learning, a server-clients architecture. The framework includes a local trajectory preprocessing and learning module for clients, a privacy-preserving clustering module, and a server-side aggregation module, where the privacy-preserving clustering module leverages differential privacy to handle data heterogeneity across clients.

See event here

Abstract

The financial services industry is experiencing a significant shift with the rise of blockchain technology and decentralized finance (DeFi), which challenges the industry’s high-cost centralized control approach. Distributed Ledger Technology (DLT), including permissionless blockchain and its novel organizational form, Decentralized Autonomous Organizations (DAOs), offer transformative potential through transparent, efficient, and decentralized financial transactions and governance structures. However, integrating these technologies into regulated financial services poses challenges due to the need to reconcile blockchain and DAOs’ decentralized nature with stringent regulatory requirements.

This dissertation, developed over a three-year PhD program, consolidates six publications into a comprehensive assessment framework for determining when DAOs and DeFi protocols are suitable for regulated financial services. It explores the effective implementation of DLT in regulated environments to reduce costs and enhance operational efficiencies. The research categorizes the publications by financial services domain, using prototyping and analysis to balance decentralization ideals with regulatory requirements. The dissertation identifies key properties of blockchain and DAOs, such as transparency, immutability, and decentralization, assessing their applicability within traditional financial systems. It offers practical tools for evaluating the suitability and compliance of DAOs in regulated financial services, mapping DAOs to organizational design theory, and highlighting their potential to reduce transaction costs and enhance trust.

Using Design Science Research (DSR) methodologies, the dissertation develops robust frameworks, applications, and taxonomies through qualitative data collection, thematic analysis, and formal methods. It provides empirical insights and tools for practitioners, including frameworks for assessing DAO governance structures and regulatory compliance. The dissertation concludes that while DAOs and DeFi protocols show promise, their application in regulated financial markets remains challenging due to regulatory complexities arising from centralized control preferences and norms.

The dissertation contributes to Information Systems (IS) research by presenting an evolution of DAOs over time and proposing an evaluation framework to assess and enhance the understanding of blockchain technology and DAOs in regulated financial services. Further, the individual publications contribute to their respective field by presenting applicable building blocks for financial services and domain-specific use cases for the DAO ecosystem.

The research provides insights and tools that deepen the understanding of DAOs’ potential and limitations in digital finance, offering a foundation for future research and practical applications. By incorporating law, economics, organizational theory, and IS perspectives, the dissertation highlights the need for new regulatory frameworks to accommodate DAOs and DLT. It suggests that while DAOs and DeFi protocols can reduce transaction costs and enhance transparency, their application in regulated finance is limited by compliance challenges and the need for complete decentralization.

The analysis suggests that ‘sufficient decentralization’ is a mirage and that complete decentralization on all three dimensions – technically, organizationally, and in service availability to the public – is the only viable option for DeFi in regulated financial markets, using the EU’s recent regulation for digital finance. While we have yet to see whether the new EU regulation and this high bar for decentralization will mean the end for DeFi and financial services DAOs in the EU, blockchain technology appears generally beneficial for finance involving untrusted parties, where intermediaries usually fulfill this role at a very high socioeconomic cost.

Supervisors

Principal Supervisor Omry Ross
Co-Supervisor Friedrich Henglein

Assessment Committee

Associate Professor Tijs Slaats, Computer Science
Professor Juho Lindman, University of Gothenburg
Assistant Professor Nina-Birte Schirrmacher, VU Amsterdam

See event here