Towards a Unified Analysis of Neural Networks in Nonparametric Instrumental Variable Regression: Optimization and Generalization

Zonghao Chen, Atsushi Nitanda, Arthur Gretton, Taiji Suzuki · 2025

Read Towards a Unified Analysis of Neural Networks in Nonparametric Instrumental Variable Regression: Optimization and Generalization by Zonghao Chen, Atsushi Nitanda, Arthur Gretton, Taiji Suzuki (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-14710.txt.

Introduction

Zonghao Chen, Atsushi Nitanda, and two co-authors prove the first global convergence result for neural networks used in instrumental variable regression. That regression style untangles cause from correlation when a hidden factor spoils direct fitting, and its standard two-stage least squares method becomes a two-level optimization once networks do the fitting. Two-level problems are hard to train and harder to certify.

The team views training through mean-field Langevin dynamics and turns the two-level problem into a single Lagrangian one. The resulting fully first-order algorithm comes with both a convergence bound and a generalization bound. The analysis exposes a trade-off in one key weight between training speed and statistical safety.

Catalog record

CREATORZonghao Chen, Atsushi Nitanda, Arthur Gretton, Taiji Suzuki
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE275 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-14710

Browse the full library · About the Tower of Babel