SLOE: A Faster Method for Statistical Inference in High-Dimensional Logistic Regression

Authors: Steve Yadlowsky, Taedong Yun, Cory Y McLean, Alexander D'Amour

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we evaluate bηSLOE in simulated data, where ground truth parameter values are known.
Researcher Affiliation Industry Steve Yadlowsky yadlowsky@google.com Taedong Yun tedyun@google.com Cory Mc Lean cym@google.com Alexander D Amour alexdamour@google.com Google Research, Brain Team Google Health
Pseudocode No The paper describes the mathematical derivation and approximation for SLOE but does not include a formal pseudocode or algorithm block.
Open Source Code Yes Code in Supplement
Open Datasets Yes The Heart Disease dataset (downloadable from the UCI Machine Learning Repository)... 136 training examples and 20 predictors (κ = 0.15).
Dataset Splits No The paper does not explicitly specify a separate 'validation' dataset split for hyperparameter tuning or model selection.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory, or cloud instance types) used for running experiments.
Software Dependencies Yes We implemented this estimator in Python, using scikit-learn to perform the MLE, and numpy / scipy [Harris et al., 2020, Virtanen et al., 2020] for the high dimensional adjustment and inference.
Experiment Setup Yes In our simulations, we use a data generating process parameterized by the sample size n, the aspect ratio κ, and signal strength γ2. We show the results of these coverage experiments, with n = 4000.