Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The paper introduces NetworkNet, a deep neural network architecture designed to model nodal heterogeneity in random networks with high-dimensional, sparse nodal attributes. NetworkNet employs a tailored neural architecture to parameterize attribute-driven heterogeneity and incorporates a scalable attribute selection mechanism. The method consistently estimates nodal expansiveness and popularity while performing data-driven attribute selection, and it is validated through simulations and an author-citation network analysis.
NetworkNet offers a statistically rigorous and scalable deep learning approach to uncover hidden patterns of influence and popularity within complex networks by simultaneously modeling nodal heterogeneity and selecting influential attributes.
Heterogeneous network data with rich nodal information become increasingly prevalent across multidisciplinary research, yet accurately modeling complex nodal heterogeneity and simultaneously selecting influential nodal attributes remains an open challenge. This problem is central to many applications in economics and sociology, when both nodal heterogeneity and high-dimensional individual characteristics highly affect network formation. We propose a statistically grounded, unified deep neural network approach for modeling nodal heterogeneity in random networks with high-dimensional nodal attributes, namely ``NetworkNet''. A key innovation of NetworkNet lies in a tailored neural architecture that explicitly parameterizes attribute-driven heterogeneity, and at the same time, embeds a scalable attribute selection mechanism. NetworkNet consistently estimates two types of latent heterogeneity functions, i.e., nodal expansiveness and popularity, while simultaneously performing data-driven attribute selection to extract influential nodal attributes. By unifying classical statistical network modeling with deep learning, NetworkNet delivers the expressive power of DNNs with methodological interpretability, algorithmic scalability, and statistical rigor with a non-asymptotic approximation error bound. Empirically, simulations demonstrate strong performance in both heterogeneity estimation and high-dimensional attribute selection. We further apply NetworkNet to a large-scale author-citation network among statisticians, revealing new insights into the dynamic evolution of research fields and scholarly impact.