ICE-BeeM: Identifiable Conditional Energy-Based Deep Models Based on Nonlinear ICA
Authors: Ilyes Khemakhem, Ricardo Monti, Diederik Kingma, Aapo Hyvarinen
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | A thorough empirical study shows that representations learned by our model from real-world image datasets are identifiable, and improve performance in transfer learning and semi-supervised learning tasks. |
| Researcher Affiliation | Collaboration | 1 Gatsby Unit, University College London 2 Google Research 3 Université Paris-Saclay, Inria 4 University of Helsinki |
| Pseudocode | No | The paper does not contain any sections explicitly labeled 'Pseudocode' or 'Algorithm', nor are there any structured code-like blocks detailing procedures. |
| Open Source Code | Yes | Code for reproducibility is available here. |
| Open Datasets | Yes | We explore the importance of identifiability and the applicability of ICE-Bee M in a series of experiments on image datasets (MNIST, Fashion MNIST, CIFAR10 and CIFAR100). |
| Dataset Splits | No | The paper describes using classes 0-7 for training and classes 8-9 as unseen classes for evaluation in transfer and semi-supervised learning tasks. However, it does not provide specific dataset split information like exact percentages, sample counts, or citations to predefined train/validation/test splits for the general datasets (MNIST, CIFAR10, CIFAR100). |
| Hardware Specification | No | The paper does not provide specific details regarding the hardware used for running the experiments, such as GPU or CPU models, memory specifications, or cloud computing instance types. |
| Software Dependencies | No | The paper mentions several software-related terms and tools in its references (e.g., PyTorch, Adam optimizer), but it does not specify exact version numbers for any libraries, frameworks, or solvers used in its implementation, which are necessary for reproducibility. |
| Experiment Setup | No | The paper describes the network architectures used in Appendix A.1 (e.g., number of layers and hidden dimensions) and mentions the training objectives (CDSM, CFCE). However, it does not provide specific hyperparameter values such as learning rate, batch size, number of epochs, or optimizer settings, which are crucial for replicating the experimental setup. |