Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Scalable Quasi-Bayesian Inference for Instrumental Variable Regression
Ziyu Wang, Yuhao Zhou, Tongzheng Ren, Jun Zhu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace bootstrap with quasi-Bayesian inference on NN model | dual IV NN model · Airline demand, N=1000 | CI coverage: .86 ± .02 -> .94 ± .01 | Improved |
| replace bootstrap with quasi-Bayesian inference on RBF kernel model | dual IV RBF kernel model · Airline demand, N=1000 | CI coverage: .45 ± .02 -> .77 ± .02 | Improved |
| replace bootstrap with quasi-Bayesian inference on NN model | dual IV NN model · Airline demand, N=1000 | test normalized MSE: .06 ± .03 -> .04 ± .00 | Improved |
| replace bootstrap with quasi-Bayesian inference on RBF kernel model | dual IV RBF kernel model · Airline demand, N=1000 | test normalized MSE: .17 ± .01 -> .17 ± .01 | No clear effect |
| replace RBF kernel with NN model in quasi-Bayesian IV | QB · Airline demand, N=1000 | test normalized MSE: .17 ± .01 -> .04 ± .00 | Improved |
| replace RBF kernel with NN model in quasi-Bayesian IV | QB · Airline demand, N=1000 | CI coverage: .77 ± .02 -> .94 ± .01 | Improved |
| use quasi-Bayesian closed-form instead of BayesIV | QB · 1D simulation run-time, N=1000 | run-time (s): 655 -> 2.37 | Improved |
| use approximate inference on GPU instead of closed-form CPU | QB · 1D simulation run-time, N=20000 | run-time (s): 1150 -> ∼140 | Improved |
| use approximate inference on GPU instead of closed-form CPU | QB · 1D simulation run-time, N=1000 | run-time (s): 2.37 -> ∼10 | Hurt |
| increase sample size from 1000 to 10000 for bootstrap | BS-NN · Airline demand | CI coverage | Improved |
| weak instrument (alpha=0.05) comparison of QB vs bootstrap | QB-RBF · 1D sine simulation | CI coverage | Improved |
Closest papers by meaning
- Nonparametric Instrumental Variable Regression through Stochastic Approximate Gradients (NeurIPS 2024 · similarity 0.89)
- Stochastic Optimization Algorithms for Instrumental Variable Regression with Streaming Data (NeurIPS 2024 · similarity 0.88)
- Optimality and Adaptivity of Deep Neural Features for Instrumental Variable Regression (ICLR 2025 · similarity 0.86)
- Demystifying Spectral Feature Learning for Instrumental Variable Regression (NeurIPS 2025 · similarity 0.85)
- Learning Decision Policies with Instrumental Variables through Double Machine Learning (ICML 2024 · similarity 0.84)
- Data-Driven Selection of Instrumental Variables for Additive Nonlinear, Constant Effects Models (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
