Up to four PhD scholarships are available for June-December 2019 start to contribute to software engineering and databases related to machine learning (ML) and cybersecurity as keys to unlock big multi-modal time-series data.

The following topics for PhD research plans are called for:

Topic 1: testing the hypothesis that multimodal time-series data can reveal early indicators of disease, details to differentiate disease subtypes, information on disease progression, and hidden knowledge about prognosis

Goal: Test the hypothesis that multimodal time-series data can reveal early indicators of disease, details to differentiate disease subtypes, information on disease progression, and hidden knowledge about prognosis by focusing on

1)  feature engineering, extraction, and selection and making the related ML more and more elegant by simplifying data sources and their feature representations for ML. This field of ML is known as dimensionality reduction and/or

2) designing, developing, and evaluating methods for ML and their evaluation.  This field of ML is known as multi-modal pattern analysis.

Topic 2: Linking the rich variety of data sources together to study the contribution of each source to the usefulness (e.g., correctness and cost of data processing) and usability (patients’ perception of trustworthiness and ease of use) of our ML system

Goal: Link the rich variety of data sources together to study the contribution of each source to the usefulness (e.g., correctness and cost of data processing) and usability (patients’ perception of trustworthiness and ease of use) of our ML system with the focus on the intersection of ML, sensor technologies, and/or record linking. We will be using two complementary strategies to solve this record linkage problem with respect to geographic location, time, and person’s identity as follows: First, a novel approach of integrating all sensing into a single device (i.e., the individual’s smartphone with appropriate accessories) that circumvents the need for record linkage. Second, linking these recordings with existing environmental and other databases in a way that is sufficiently accurate, secure, and ethical (e.g., data subject’s informed consent for data gathering and use).

Topic 3: Building cyber-physical security into this system and assessing it as trustworthy and compliant with the relevant standards and constraints

Goal: Build cyber-physical security into this system and assess it as trustworthy and compliant with the relevant standards and constraints by focusing on

1) technical aspects in designing, developing, and evaluating the proposed system and/or

2) socio-technical and information systems management aspects.

The device proposed in this project is, from the information collection and exchange perspective, is a cyber-physical system (CPS). This CPS is composed of both hardware components, such as sensors, actuators, and embedded systems, and software products. Notably, the heterogeneity of hardware and software components have introduced significant difficulties to security and privacy protection of the CPS. With the complex cyber-physical interactions, threats and vulnerabilities becomes difficult to assess, and new security issues arise. Also, it becomes difficult to identify, trace and examine the attacks, which may originate from, move between, and target at multiple CPS components.