Identifiability of Label Noise Transition Matrix
Authors: Yang Liu, Hao Cheng, Kun Zhang
ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 6. Empirical Evidence: Disentangled Features ... The overall experiments are shown in Table 1. |
| Researcher Affiliation | Collaboration | Yang Liu 1 2 Hao Cheng 1 Kun Zhang 3 4 ... 1University of California, Santa Cruz 2Byte Dance Research 3Carnegie Mellon University 4Mohamed bin Zayed University of Artificial Intelligence. |
| Pseudocode | Yes | Algorithm 1 Key Steps of HOC |
| Open Source Code | Yes | Code is available at https://github.com/ UCSC-REAL/Identifiability. |
| Open Datasets | Yes | Note all the three encoders are trained on CIFAR100 datasets and generate feature for CIFAR10 to estimate noise transition matrix. |
| Dataset Splits | No | The paper mentions training on datasets like CIFAR100 and CIFAR10 but does not provide specific details on how the data was split into training, validation, and test sets, beyond implying the use of 'noisy dataset' or 'randomly selected set of instances'. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models or types) used for conducting the experiments. |
| Software Dependencies | No | The paper mentions software components such as 'optimizer: Adam', 'SGD', and refers to 'official codebase of IPIRM' which implies Python/PyTorch, but it does not specify any version numbers for these software dependencies. |
| Experiment Setup | Yes | The hyper-parameters for estimating transition matrix are consistent with official implementation of HOC: optimizer: Adam, learning rate: 0.1, number of iterations: 1500. |