Regression Models for Reciprocity in Directed Graphs
Prof. Chenlei LENG
Chair Professor of Statistics and Machine Learning
Department of Applied Mathematics
The Hong Kong of Polytechnic University
Date: 21 November 2025 (Friday)
Time: 10:30-12:00
Venue: E22-G015
Host: Prof. Degui LI, Distinguished Professor in Business Economics
Abstract
Reciprocity—the tendency for directed edges to appear in mutual pairs—is a key feature of many networks but remains difficult to model, especially in sparse settings. This talk introduces two regression frameworks for capturing reciprocity in directed graphs with covariates. The first uses a novel Bernoulli formulation to distinguish reciprocal from non-reciprocal ties, along with an inference method and analysis of effective sample sizes under sparsity. The second extends the classical p1 model by allowing for link-specific reciprocity alongside node heterogeneity. We propose a new estimator based on a conditioning argument and provide its minimax optimality. Numerical examples illustrate the models’ performance and support the theory. This is based on joint work with Rui Feng.
Speaker
Prof. Chenlei LENG is Chair Professor of Statistics and Machine Learning in the Department of Applied Mathematics at the Hong Kong Polytechnic University. His research develops new statistical and machine learning methods for analyzing large, complex datasets, with particular focus on high-dimensional, correlated, and network-structured data.
Professor LENG is a Fellow of the Institute of Mathematical Statistics and an Elected Member of the International Statistical Institute. His leadership roles include chairing the Research Section of the Royal Statistical Society, co-directing the Oxford–Warwick Centre for Doctoral Training, and serving as an inaugural Turing Fellow at the Alan Turing Institute.
All are welcome!
