Latent Position

Latent position modeling aims to uncover hidden structural relationships within data, often represented as networks or graphs, by assigning positions to nodes in a lower-dimensional space. Current research focuses on improving the accuracy and interpretability of these positions, exploring methods like self-attention variance in transformer models and employing algorithms such as the Nadaraya-Watson estimator and Bradley-Terry models for position estimation. These advancements have applications in diverse fields, including natural language processing, social network analysis (e.g., measuring political polarization), and graph neural networks, offering valuable insights into complex systems and facilitating more effective data analysis.

Papers

October 29, 2024

Node Regression on Latent Position Random Graphs via Local Averaging
Martin Gjorgjevski, Nicolas Keriven, Simon Barthelmé, Yohann De Castro
Latent Graph Graph Regression Self Averaging Nadaraya Watson Node Level Regression Latent Position

May 23, 2023

Latent Positional Information is in the Self-Attention Variance of Transformer Language Models Without Positional Embeddings
Ta-Chung Chi, Ting-Han Fan, Li-Wei Chen, Alexander I. Rudnicky, Peter J. Ramadge
Self Attention Transformer Language Model Positional Embeddings Frozen Language Model Latent Position

March 30, 2023

The Graphical Nadaraya-Watson Estimator on Latent Position Models
M. Gjorgjevski
Graph Neural Network Graph Drawing Mean Estimation Nadaraya Watson Latent Position

March 21, 2023

Large Language Models Can Be Used to Estimate the Latent Positions of Politicians
Patrick Y. Wu, Jonathan Nagler, Joshua A. Tucker, Solomon Messing
Generative Large Language Model Left Leaning United State Politician Latent Position

January 23, 2023

Characterizing Polarization in Social Networks using the Signed Relational Latent Distance Model
Nikolaos Nakis, Abdulkadir Çelikkanat, Louis Boucherie, Christian Djurhuus, Felix Burmester, Daniel Mathias Holmelund, Monika Frolcová, Morten Mørup
Graph Representation Learning Social Network Political Polarization Latent dIstance Model Latent Position

Latent Position

Papers

Node Regression on Latent Position Random Graphs via Local Averaging

Latent Positional Information is in the Self-Attention Variance of Transformer Language Models Without Positional Embeddings

The Graphical Nadaraya-Watson Estimator on Latent Position Models

Large Language Models Can Be Used to Estimate the Latent Positions of Politicians

Characterizing Polarization in Social Networks using the Signed Relational Latent Distance Model