Outside the Echo Chamber: Optimizing the Performative Risk

Authors: John P Miller, Juan C Perdomo, Tijana Zrnic

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

Reproducibility Variable Result LLM Response
Research Type Experimental We complement our theoretical findings with an empirical evaluation of different methods on two tasks: the strategic classification simulator from Perdomo et al. (2020), and a synthetic linear regression example. ... For each of the following experiments, we run each algorithm 50 times and display 95% bootstrap confidence intervals.
Researcher Affiliation Academia 1University of California, Berkeley. Correspondence to: John Miller <miller john@berkeley.edu>, Juan C. Perdomo <jcperdomo@berkeley.edu>, Tijana Zrnic <tijana.zrnic@berkeley.edu>.
Pseudocode Yes Algorithm 1 Two-Stage Algorithm for Location Families
Open Source Code No The paper does not include an unambiguous statement that the authors are releasing the code for the work described in this paper, nor does it provide a direct link to a source-code repository.
Open Datasets Yes We complement our theoretical findings with an empirical evaluation of different methods on two tasks: the strategic classification simulator from Perdomo et al. (2020), and a synthetic linear regression example.
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4) needed to replicate the experiment.
Experiment Setup Yes We provide a formal description of all the procedures, as well as a detailed description of the experimental setup in Appendix D. Appendix D itself details parameters like 'x ~ N(0, Σx), U_y ~ N(0, σ^2_y), y = β^T x + µ^T θ + U_y' for linear regression, and 'add ℓ2-regularization to the logistic loss' for strategic classification.