About Me

I am a physicist interested in the theory of complex networks as well as their applications to urban and social systems. My research draws on ideas from a range of disciplines including statistical physics, information theory, Bayesian inference, scientific computing, machine learning, urban science, and geography.

Currently I am an Assistant Professor at the University of Hong Kong (HKU), jointly appointed in the Institute of Data Science and Department of Urban Planning and Design. I received my PhD in Physics at the University of Michigan in 2021 under the supervision of Mark Newman, and received undergraduate degrees in Physics and Mathematics at the University of Rochester in my hometown of Rochester, New York.

My main research interests are:

1. Developing statistically principled unsupervised learning methods for noisy network data

Networks present new challenges for inference and learning due to their relational, hierarchical, and high (usually infinite!) dimensional nature. There are also many ways in which uncertainty can be introduced into network datasets: Networks may be noisy due to statistical or systematic measurement errors; Networks measured in cross-sectional or longitudinal studies may be subject to errors across different experimental environments; Samples of networks or their partitions from a Bayesian posterior distribution are subject to the intrinsic uncertainties in their corresponding statistical model. In this line of research I aim to develop principled, flexible, and interpretable methods that do not overfit or rely on ad hoc heuristics to model network data in a variety of forms (temporal/multilayer networks, hypergraphs, etc). The methods I develop typically require the formulation of new learning objectives as well as novel sampling and optimization techniques to reveal the patterns of interest and distinguish statistically meaningful structure from noise.

• Example publications:
• Compressing network populations with modal networks reveals structural diversity
• Implicit models, latent compression, intrinsic biases, and cheap lunches in community detection
• Inference of dynamic hypergraph representations in temporal interaction data
• Identifying hubs in directed networks
• Clustering of heterogeneous populations of networks

2. Improving the efficiency and interpretability of network model fitting and evaluation

The discrete, high-dimensional nature of networks also introduces new challenges for model optimization and evaluation, since many objectives of interest are combinatorial in nature, depend on large-scale network properties such as long loops or paths, and produce unwieldy outputs in the form of sets or partitions. As a result, many existing methods for network inference and model evaluation suffer from issues of scalability, interpretability, and subtle but important biases that can compromise the quality of the results in practical applications. In this line of research I aim to develop improved computational methods to enable the scaling up of inference methods to large networks, as well as construct robust, interpretable measures for comparing network structures derived from first principles.

• Example publications:
• Belief propagation for networks with loops
• Representative community divisions of networks
• Heterogeneous message passing for heterogeneous networks
• Normalized mutual information is a biased measure for classification and community detection

3. Modelling urban mobility networks and their impact on city prosperity and resilience

Urban mobility infrastructure and dynamics play a critical role in determining the social, economic, and physical well-being of cities and their residents. Structural networks such as street and other transportation networks facilitate the distribution of resources as well as the accessibility of jobs, amenities, and services. Meanwhile, networks of human mobility dynamics determine the mixing and contact patterns of urban residents, which impacts the spread of ideas and infectious diseases as well as the vibrancy, safety, and congestion of urban spaces. My research aims to apply insights from network science and statistical physics to understand how these structural and dynamical mobility networks impact the propserity and resilience of global cities.

• Example publications:
• From the betweenness centrality in street networks to structural invariants in random planar graphs
• Connecting intercity mobility with urban welfare
• Integrating online and offline data for crisis management: Online geolocalized emotion, policy response, and local mobility during the COVID crisis
• Impact of urban structure on infectious disease spreading

4. Characterizing heterogeneity and correlations in urban spatial data with network methods

There is an abundance of spatial data available to help understand the structure and function of urban regions, from census data providing demographic information for urban populations, to the precise locations of points of interest such as services or attractions, to satellite imagery capturing large scale land usage patterns. Due to high levels of heterogeneity in population density and the strong dependence of statistical analyses on spatial scale, topologically motivated methods based on spatial network analysis provide a robust lens through which to study urban spatial data. My research aims to develop new methods fusing network science and information theory to quantitaively characterize urban spatial data, with the goal of providing a set of tools that can be used to understand spatially embedded phenomena such as socioeconomic inequality and segregation.

• Example publications:
• Spatial regionalization based on optimal information compression
• Characterizing network circuity among heterogeneous urban amenities
• Information theoretic network approach to socioeconomic correlations

For a complete list of publications, check out my Publications page. All papers should be either open access or freely available on arXiv. Feel free to get in touch if you have trouble accessing any of the papers.

I am looking for researchers at all levels (Postdoctoral Fellows, PhD students, and Research Assistants) to work on the lines of research listed above. Please send me your CV via email if you are interested. I encourage you to include in your email a paragraph or so stating your research interests/experience and how you feel these interests fit with the research areas outlined above, as this will better facilitate matching you to potential projects. PhD applicants may also be interested in checking out the Hong Kong PhD Fellowship Scheme, which provides an excellent stipend and support for PhD studies in Hong Kong.

I make sure to read all applicant emails, but do not take it personally if I forget to reply since I often won’t be able to process these emails immediately. In general I am always happy to discuss opportunities for supervision, but take note that openings for positions may be subject to university- and department-level admissions criteria as well as project funding. I will try my best to be clear about these sources of variability and how they apply to each interested candidate. I very much appreciate your interest in my research!

Group members

• Sebastian Morel-Balbi, Postdoctoral Fellow (2023-).

Teaching

• Statistical Inference and Machine Learning for Network Data (DATA 8002, Data Science PhD/MPhil Program, Fall 2023).
• Transport Network Analysis and Modelling (MUDT 5010, Urban Design and Transport MSc Program, Fall 2023).
• Science of Cities (URBA 6006, Urban Analytics MSc Program, Fall 2023).