Document Type


Date of Award


Degree Name

Doctor of Philosophy in Business Data Science - (Ph.D.)


Data Science

First Advisor

Dantong Yu

Second Advisor

Ioannis Koutis

Third Advisor

Baruch Schieber

Fourth Advisor

Yi Chen

Fifth Advisor

Junmin Shi


The world has never been more connected, led by the information technology revolution in the past decades that has fundamentally changed the way people interact with each other using social networks. Consequently, enormous human activity data are collected from the business world and machine learning techniques are widely adopted to aid our decision processes. Despite of the success of machine learning in various application scenarios, there are still many questions that need to be well answered, such as optimizing machine learning outcomes when desired knowledge cannot be extracted from the available data. This naturally drives us to ponder if one can leverage some side information to populate the knowledge domain of their interest, such that the problems within that knowledge domain can be better tackled.

In this work, such problems are investigated and practical solutions are proposed. To leverage machine learning in any decision-making process, one must convert the given knowledge (for example, natural language, unstructured text) into representation vectors that can be understood and processed by machine learning model in their compatible language and data format. The frequently encountered difficulty is, however, the given knowledge is not rich or reliable enough in the first place. In such cases, one seeks to fuse side information from a separate domain to mitigate the gap between good representation learning and the scarce knowledge in the domain of interest. This approach is named Cross-Domain Knowledge Transfer. It is crucial to study the problem because of the commonality of scarce knowledge in many scenarios, from online healthcare platform analyses to financial market risk quantification, leaving an obstacle in front of us benefiting from automated decision making. From the machine learning perspective, the paradigm of semi-supervised learning takes advantage of large amount of data without ground truth and achieves impressive learning performance improvement. It is adopted in this dissertation for cross-domain knowledge transfer.

Furthermore, graph learning techniques are indispensable given that networks commonly exist in real word, such as taxonomy networks and scholarly article citation networks. These networks contain additional useful knowledge and are ought to be incorporated in the learning process, which serve as an important lever in solving the problem of cross-domain knowledge transfer. This dissertation proposes graph-based learning solutions and demonstrates their practical usage via empirical studies on real-world applications. Another line of effort in this work lies in leveraging the rich capacity of neural networks to improve the learning outcomes, as we are in the era of big data.

In contrast to many Graph Neural Networks that directly iterate on the graph adjacency to approximate graph convolution filters, this work also proposes an efficient Eigenvalue learning method that directly optimizes the graph convolution in the spectral space. This work articulates the importance of network spectrum and provides detailed analyses on the spectral properties in the proposed EigenLearn method, which well aligns with a series of CNN models that attempt to have meaningful spectral interpretation in designing graph neural networks. The disser-tation also addresses the efficiency, which can be categorized in two folds. First, by adopting approximate solutions it mitigates the complexity concerns for graph related algorithms, which are naturally quadratic in most cases and do not scale to large datasets. Second, it mitigates the storage and computation overhead in deep neural network, such that they can be deployed on many light-weight devices and significantly broaden the applicability. Finally, the dissertation is concluded by future endeavors.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.