Learning from Aggregate Observations
Authors: Yivan Zhang, Nontawat Charoenphakdee, Zhenguo Wu, Masashi Sugiyama
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on three problem settings classification via triplet comparison and regression via mean/rank observation indicate the effectiveness of the proposed method. (Abstract) and In this section, we present the empirical results of the proposed method. All experimental details can be found in Appendix E. More extensive experiments on 20 regression datasets and 30 classification datasets are provided in Appendix F. (Section 6) |
| Researcher Affiliation | Academia | Yivan Zhang The University of Tokyo / RIKEN, Nontawat Charoenphakdee The University of Tokyo / RIKEN, Zhenguo Wu The University of Tokyo, Masashi Sugiyama RIKEN / The University of Tokyo |
| Pseudocode | No | The paper describes mathematical formulations and theoretical derivations, but it does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code for the described methodology, nor does it provide a direct link to a code repository. |
| Open Datasets | Yes | We evaluate our method on three image classification datasets, namely MNIST,4 Fashion-MNIST (FMNIST),5 and Kuzushiji-MNIST (KMNIST),6... We conducted experiments on 6 UCI benchmark datasets7 with footnotes linking to the sources: 4MNIST [Le Cun et al., 1998] http://yann.lecun.com/exdb/mnist/, 5Fashion-MNIST [Xiao et al., 2017] https://github.com/zalandoresearch/fashion-mnist, 6Kuzushiji-MNIST [Clanuwat et al., 2018] http://codh.rois.ac.jp/kmnist/, 7UCI Machine Learning Repository [Dua and Graff, 2017] https://archive.ics.uci.edu. |
| Dataset Splits | Yes | The training set is further divided into training and validation sets with a ratio of 9:1. We also randomly selected 5% of training data for optimal permutation for modified accuracy calculation and 5% for visualization. (Appendix E.1) and All datasets are split into training, validation, and testing sets using a 8:1:1 ratio. (Appendix F.1.1) |
| Hardware Specification | No | The paper mentions 'the Reedbush supercomputer system' in the Acknowledgement section, but it does not provide specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running experiments. |
| Software Dependencies | No | The paper states 'We used PyTorch for implementation.' (Appendix E.1) and 'We used LightGBM [Ke et al., 2017] as the GBM model with default parameters.' (Appendix E.2), but it does not specify version numbers for these software components. |
| Experiment Setup | Yes | We used Adam [Kingma and Ba, 2014] as the optimizer, with a learning rate of 1e-3, beta1=0.9, beta2=0.999, and a batch size of 256. (Appendix E.1) and We used Adam [Kingma and Ba, 2014] as the optimizer, with a learning rate of 1e-2, beta1=0.9, beta2=0.999, and a batch size of 128. (Appendix E.2) |