A Statistical Online Inference Approach in Averaged Stochastic Approximation

Authors: Chuhan Xie, Zhihua Zhang

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct experiments to perform inference with both random scaling and other traditional inference methods, and finds that the former has a more accurate and robust performance. In this section we conduct the empirical analysis of the random scaling method via Monte Carlo experiments.
Researcher Affiliation Academia Chuhan Xie School of Mathematical Sciences Peking University Beijing, 100871, China ch_xie@pku.edu.cn Zhihua Zhang School of Mathematical Sciences Peking University Beijing, 100871, China zhzhang@math.pku.edu.cn
Pseudocode No The paper uses mathematical equations to describe algorithms (e.g., SA procedure, SGD updates, Q-learning), but it does not contain any structured pseudocode or algorithm blocks that are clearly labeled as such.
Open Source Code Yes The implementation code is available in https://github.com/bangoz/sa-inference.
Open Datasets No The data are generated from yt = x t β + εt, t ≥ 1, where xt is a d-dimensional covariate following the multivariate normal distribution N(0, Id), εt is the noise from N(0, 1), and β is equi-spaced on the interval [0, 1]. The data are generated from P(yt = 1) = 1 / (1 + exp(-x t β )), P(yt = 0) = exp(-x t β ) / (1 + exp(-x t β )), t ≥ 1. The paper uses synthetically generated data for its experiments and does not provide access information for a publicly available dataset.
Dataset Splits No The paper states: "The simulation results are based on 1,000 replications." However, it does not specify explicit training/validation/test dataset splits, as the data is generated for each simulation rather than being drawn from a fixed dataset with predefined splits.
Hardware Specification Yes All experiments are run on Intel Xeon 14-core CPUs.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., library names with their corresponding versions) that are needed to replicate the experiment.
Experiment Setup Yes The dimension of x is set as d = 5, 20. The constant learning rate is set as η = 0.01, 0.05, 0.1. The initial value of β0 is set as zero. The simulation results are based on 1,000 replications.