Yanning SHEN

We live in an era of data deluge. Pervasive media collect massive amounts of data in a wide variety of formats, constructing networks of a wide range of physical, biological, and social interdependencies. Learning from these large volumes of network data can bring significant science and technology advances.

As an assistant professor at the University of California, Irvine, Yanning Shen focuses on algorithms, analysis, and application of machine learning, optimization, and statistical signal processing tools to data science and network science. Her research includes scalable nonlinear and tensor-based learning from high-dimensional (network) data, which finds exciting applications in understanding the structure and dynamics of social-, biological-, financial-, and engineering systems.

To achieve scalable and adaptive learning for big streaming data, Yanning developed a new framework with theoretical performance guarantees to efficiently track the latent low-dimensional structures from incomplete and corrupt datasets that are typically encountered in practice. She subsequently investigated nonlinear functional learning and developed a data-driven learning scheme that showcased its effectiveness in environments with unknown dynamics.

Yanning also advocated novel kernel-based graph topology inference approaches that could account for the nonlinear dependencies among nodes and across time. To facilitate real-time operation and inference of time-varying networks, she further developed an adaptive tensor decomposition-based scheme, which tracks the topology-revealing tensor factors.

With insights gained from the above research efforts, Yanning further envisions broadening the theoretical and algorithmic framework to online scalable machine learning over networks. Her recent research results have revealed the benefits of considering the underlying network topologies for classification and clustering, as well as the dimensionality reduction of data observed over graphs.

Yanning’s long-term research plan is to promote trustworthy intelligence over interconnected systems. She plans to develop a theoretical and algorithmic foundation for addressing fairness, privacy, and interpretability in machine learning over inter-connected systems that can be modeled using graphs.