$H$-Consistency Guarantees for Regression
Authors: Anqi Mao, Mehryar Mohri, Yutao Zhong
ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We report favorable experimental results in Section 6. In this section, we demonstrate empirically the effectiveness of the smooth adversarial regression algorithms introduced in the previous section. |
| Researcher Affiliation | Collaboration | 1Courant Institute of Mathematical Sciences, New York, NY; 2Google Research, New York, NY. |
| Pseudocode | No | The paper describes methods and theoretical derivations but does not include any pseudocode or explicitly labeled algorithm blocks. |
| Open Source Code | No | The paper does not provide an explicit statement about releasing its source code or a link to a code repository for the methodology described. |
| Open Datasets | Yes | We studied two real-world datasets: the Diabetes dataset (Efron et al., 2004) and the Diverse MAGIC wheat dataset (Scott et al., 2021) |
| Dataset Splits | No | The paper mentions training and testing but does not explicitly provide details about specific training/validation/test dataset splits, proportions, or cross-validation setup. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions the 'CVXPY library (Diamond & Boyd, 2016)' but does not specify a version number for the library itself. |
| Experiment Setup | Yes | For our smooth adversarial regression losses (2), we chose L = ℓ2, the squared loss, and L = ℓδ with δ = 0.2, the Huber loss, setting τ = 1 as the default. Other choices for the regression loss functions and the value of τ may yield better performance, which can typically be selected by cross-validation in practice. Both our smooth adversarial regression losses and the adversarial squared loss were optimized using the CVXPY library (Diamond & Boyd, 2016). |