Invariance Learning based on Label Hierarchy
Authors: Shoji Toyota, Kenji Fukumizu
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The effectiveness of the proposed framework, including the cross-validation, is demonstrated empirically. Theoretical analysis reveals that our framework can estimate the desirable invariant predictor with a hyperparameter fixed correctly, and that such a preferable hyperparameter is chosen by the proposed CV methods under some conditions. We study the effectiveness of the proposed framework and CVs through experiments, comparing them with several existing methods: empirical risk minimization (ERM), fine-tuning methods, and deep domain adaptation strategies. |
| Researcher Affiliation | Academia | Shoji Toyota The Graduate University for Advanced Studies Tokyo 190-8562, Japan shoji@ism.ac.jp Kenji Fukumizu The Institute of Statistical Mathematics The Graduate University for Advanced Studies Tokyo 190-8562, Japan fukumizu@ism.ac.jp |
| Pseudocode | Yes | Algorithm 1 CV methods. If CORRECTION = True, λ is selected by method II and if False, I. |
| Open Source Code | Yes | The code is available in Supplementary Material. |
| Open Datasets | Yes | Colored MNIST We apply our framework to Colored MNIST [10] with Y = [10] and Z := [2]. Image Net To see the performance of the proposed methods for more practical data, they are applied to the Image Net [53] with its label reannotated imitating BREEDS [52]. [53] J. Deng, A. Berg, S. Satheesh, H. Su, A. Khosla, L. Fei-Fei. Imagenet: A large-scale hierarchical image database. In CVPR, 2009. |
| Dataset Splits | Yes | We propose two methods of cross-validation (CV) for hyperparameter selection in our new IL framework. Algorithm 1 CV methods. If CORRECTION = True, λ is selected by method II and if False, I. Require: : Split De , De1 ad, ..., Den ad into K parts. |
| Hardware Specification | Yes | All experiments were performed on a machine with NVIDIA Tesla V100 GPU. |
| Software Dependencies | Yes | All our code is written using PyTorch 1.10.1. |
| Experiment Setup | Yes | Setting the maximum epoch 500 and λbefore := 1.0, we select (t, λafter) from 4 7 candidates with t {0, 100, 200, 300}, λafter {100, 101, ..., 106} by each of the CVs. We use Adam [37] with β1 = 0.5, β2 = 0.9, and learning rate 0.0001, and decay learning rate by a factor of 0.1 at 80% of total epochs. |