Functional Indirection Neural Estimator for Better Out-of-distribution Generalization
Authors: Kha Pham, Thai Hung Le, Man Ngo, Truyen Tran
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate empirically that FINE can strongly improve out-of-distribution generalization on IQ tasks that involve geometric transformations. In particular, we train FINE and competing models on IQ tasks using images from the MNIST, Omniglot and CIFAR100 datasets and test on tasks with unseen image classes from one or different datasets and unseen transformation rules. FINE not only achieves the best performance on all tasks but also is able to adapt to small-scale data scenarios. |
| Researcher Affiliation | Academia | Kha Pham1 Hung Le1 Man Ngo2 Truyen Tran1 1 Applied Artificial Intelligence Institute, Deakin University 2 Faculty of Mathematics and Computer Science, VNUHCM-University of Science |
| Pseudocode | No | The paper describes the steps of the FINE architecture and its memory reading/update processes but does not include a formally structured pseudocode block or algorithm figure. |
| Open Source Code | Yes | Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] see Supplementary [...] We include codes for our tasks and models. |
| Open Datasets | Yes | We generate data for IQ tasks described in Section 2 using images from Omniglot dataset [5], which includes 1,623 handwritten characters, and real-image CIFAR100 dataset [17]. |
| Dataset Splits | No | The paper specifies training and testing set sizes/classes (e.g., "The train and test set size is 10,000 and 20,000, respectively." for Omniglot), but it does not mention a distinct validation split. |
| Hardware Specification | No | Experiments are conducted using Py Torch on a single GPU with Adam optimizer. [...] Did you include the total amount of compute and the type of resources used (e.g., type of GPUs, internal cluster, or cloud provider)? [Yes] see Section 4. (Section 4 only mentions 'single GPU' without further specifics). |
| Software Dependencies | No | Experiments are conducted using Py Torch on a single GPU with Adam optimizer. (No version numbers are provided for PyTorch or any other software components used in the experiments). |
| Experiment Setup | Yes | For FINE, we use 4 NICE layers with 48 memories for each pair of NICE layers. [...] If not specified, models are trained with p4-CNN encoder [7]. Experiments are conducted using Py Torch on a single GPU with Adam optimizer. |