Meta-Learning Requires Meta-Augmentation
Authors: Janarthanan Rajendran, Alexander Irpan, Eric Jang
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate that meta-augmentation produces large complementary benefits to recently proposed meta-regularization techniques. (Abstract) and Finally, we show the importance of meta-augmentation on a variety of benchmarks and meta-learning algorithms. (Section 1, Introduction) and 5 Experiments (Section 5). |
| Researcher Affiliation | Collaboration | Janarthanan Rajendran University of Michigan rjana@umich.edu Alex Irpan Google Brain alexirpan@google.com Eric Jang Google Brain ejang@google.com |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code and data available at https://github.com/google-research/google-research/tree/master/meta_augmentation. |
| Open Datasets | Yes | Few shot classification benchmarks such as Mini-Image Net [36]... Omniglot The Omniglot dataset [20]... Pascal3D Pose Regression We show that for the regression problem introduced by Yin et al. [37]... |
| Dataset Splits | No | The paper mentions 'meta-training and meta-test sets of tasks' and 'Val pre-update' and 'Val post-update' in figures, implying the use of a validation set for early stopping, but it does not specify explicit percentages or sample counts for the dataset splits. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment. |
| Experiment Setup | No | The paper mentions experimental settings like '1-shot, 5-way classification' and refers to 'learning rate' and 'weight decay' as parameters, but it does not provide a comprehensive list of specific hyperparameter values or detailed system-level training configurations in the main text. |