Minimax Estimation of Conditional Moment Models

Authors: Nishanth Dikkala, Greg Lewis, Lester Mackey, Vasilis Syrgkanis

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conclude with an extensive experimental analysis of the proposed methods. and 9 Experimental Analysis
Researcher Affiliation Collaboration Nishanth Dikkala MIT nishanthd@csail.mit.edu Greg Lewis Microsoft Research glewis@microsoft.com Lester Mackey Microsoft Research lmackey@microsoft.com Vasilis Syrgkanis Microsoft Research vasy@microsoft.com
Pseudocode Yes Theorem 4. Consider the algorithm where for t = 1, . . . , T: let , ft = Oracle F (z1:n, ut i = 1{ft(zi) > 0}, wt i = |ft(zi)| ht = Oracle H
Open Source Code Yes Associated code can be found in https://github.com/microsoft/Adversarial GMM.
Open Datasets Yes MNIST dataset consisting of grayscale images of 28 28 pixels.
Dataset Splits No The paper does not explicitly state specific training, validation, or test dataset splits (e.g., percentages or sample counts) needed for reproduction.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper mentions software components like 'Elastic Net CV' and 'neural networks' but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes We consider the following data generating processes: for nx = 1 and nz 1 y = h0(x[0]) + e + δ, δ N(0, .1) x = γ z[0] + (1 γ) e + γ, z N(0, 2 Inz), e N(0, 2), γ N(0, .1) ... We consider several functional forms for h0 including absolute value, sigmoid and sin functions... We consider as classic benchmarks 2SLS with a polynomial features of degree 3 (2SLS) and a regularized version of 2SLS where Elastic Net CV is used in both stages (Reg2SLS).