Research map / Probabilistic and scientific modeling / Treatment effects
Deep IV: A Flexible Approach for Counterfactual Prediction
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use upper-bound (Jensen) loss with a single sample instead of unbiased two-sample gradient (ablation) | DeepIV second stage · Image task (MNIST-based), 20000 training points | out-of-sample counterfactual MSE: 0.32 vs 0.39-0.50 for unbiased variants; differences not statistically significant | No clear effect |
| Replace the unbiased loss with the cheaper upper bound loss (ablation) | DeepIV second stage · Image task (MNIST-based) | out-of-sample counterfactual MSE | No clear effect |
| Use causal validation loss (second-stage held-out loss) for hyperparameter selection | DeepIV · Image task (MNIST-based) | counterfactual test loss | Helped |
| Replace the standard feed-forward network with DeepIV (IV-based loss with instruments) | Simulated airline pricing (low-dimensional) · Simulation, out-of-sample structural MSE | structural MSE | Helped |
| Use DeepIV on real Bing advertiser data with text features and experiment-ID instruments instead of naive regression | Search-advertisement position effect · Bing click data, ATE click drop position 1 to 2 | ATE drop in click rate: 70% naive vs 12% causal | Helped |
Closest papers by meaning
- Adversarially Balanced Representation for Continuous Treatment Effect Estimation (AAAI 2024 · similarity 0.87)
- Distribution-Conditioned Adversarial Variational Autoencoder for Valid Instrumental Variable Generation (AAAI 2024 · similarity 0.86)
- A Dual-module Framework for Counterfactual Estimation over Time (ICML 2024 · similarity 0.86)
- Continuous Treatment Effect Estimation Using Gradient Interpolation and Kernel Smoothing (AAAI 2024 · similarity 0.85)
- Counterfactual Generative Modeling with Variational Causal Inference (ICLR 2025 · similarity 0.85)
- Quantifying Aleatoric Uncertainty of the Treatment Effect: A Novel Orthogonal Learner (NeurIPS 2024 · similarity 0.85)
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.