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 [1].
Self-Supervised Fair Representation Learning without Demographics
Authors: Junyi Chai, Xiaoqian Wang
NeurIPS 2022 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results show that our proposed method achieves better or comparable performance than state-of-the-art methods on three datasets in terms of accuracy and several fairness metrics.4 Experiments |
| Researcher Affiliation | Academia | Junyi Chai, Xiaoqian Wang Elmore Family School of Electrical and Computer Engineering Purdue University West Lafayette, IN 47906 EMAIL |
| Pseudocode | Yes | Algorithm 1 Optimization Algorithm |
| Open Source Code | Yes | Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] |
| Open Datasets | Yes | Celeb A (Liu et al., 2015): The dataset contains...Adult (Dua and Graff, 2017) : The dataset contains...COMPAS (Larson et al., 2016): The dataset contains... |
| Dataset Splits | Yes | We repeat experiments on each dataset three times and report the average results and in each repetition we randomly spilt data into 64% training data, 16% validation data and 20% test data. |
| Hardware Specification | Yes | We implement our method in Py Torch 1.10.1 with one NVIDIA RTX-3090 GPU. |
| Software Dependencies | Yes | We implement our method in Py Torch 1.10.1 with one NVIDIA RTX-3090 GPU. |
| Experiment Setup | Yes | All hyperparameters are tuned to find the best validation accuracy. The values of hyperparameter in our method are set by performing cross-validation on training data in the value range of 0.1 to 10. The hyperparameters for the comparing methods are tuned as suggested in the original paper (Hardt et al., 2016; Hashimoto et al., 2018; Hwang et al., 2020). |