Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Reward-oriented Causal Representation Learning
Authors: Zirui Yan, Emre Acartürk, Ali Tajer
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we evaluate the empirical performance of RO-CRL. We report the regret of ROCRL under both soft and hard interventions. Additional experiments, including CRL recovery, scaling behavior, comparison to baselines, and assumption violations, are deferred to Appendix C. |
| Researcher Affiliation | Academia | Zirui Yan Rensselaer Polytechnic Institute EMAIL Emre Acartürk Rensselaer Polytechnic Institute EMAIL Ali Tajer Rensselaer Polytechnic Institute EMAIL |
| Pseudocode | Yes | The pseudocode is provided in Algorithm 1 in Appendix A. |
| Open Source Code | Yes | The codebase for the experiments can be found at https://github.com/ZiruiYan/RO-CRL. |
| Open Datasets | No | This paper is purely theoretical, and the experiments only use synthetic data, so we are not aware of any negative societal impacts. |
| Dataset Splits | No | We generate a random acyclic graph on n nodes by enforcing a strictly lower-triangular weight matrix. ... We set n = 5 for these experiments and repeat the experiments 50 times. |
| Hardware Specification | Yes | Question: For each experiment, does the paper provide sufficient information on the computer resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [Yes] Justification: The experiments only need CPUs. |
| Software Dependencies | No | Portions of the publicly available code of [3, 4], available under Apache 2.0 license, is adopted in the code of our experiments. |
| Experiment Setup | Yes | Specifically, observational weights are drawn from [0.25, 1] and interventional weights are set 0.1 times the corresponding observational weights for soft interventions and to 0 for hard interventions. Noise terms are sampled i.i.d. from the uniform distribution U[0, 1]. We set n = 5 for these experiments and repeat the experiments 50 times. |