Meta-Learning with Adaptive Hyperparameters

Authors: Sungyong Baik, Myungsub Choi, Janghoon Choi, Heewon Kim, Kyoung Mu Lee

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we demonstrate the effectiveness of our proposed weight-update rule (ALFA) in few-shot learning. The experimental results validate that the Adaptive Learning of hyperparameters for Fast Adaptation (ALFA) is the equally important ingredient that was often neglected in the recent few-shot learning approaches.
Researcher Affiliation Academia Sungyong Baik Myungsub Choi Janghoon Choi Heewon Kim Kyoung Mu Lee ASRI, Department of ECE, Seoul National University {dsybaik, cms6539, ultio791, ghimhw, kyoungmu}@snu.ac.kr
Pseudocode Yes Algorithm 1 Adaptive Learning of Hyperparameters for Fast Adaptation (ALFA)
Open Source Code Yes 1The code is available at https://github.com/baiksung/ALFA
Open Datasets Yes For few-shot classification, we use the two most popular datasets: mini Image Net [38] and tiered Image Net [29]. CUB-200-2011 (denoted as CUB) [40], Triantafillou et al. [37] recently introduced a large-scale dataset, named Meta-Dataset
Dataset Splits Yes mini Image Net...is divided into 3 subsets of classes without overlap: 64 classes for meta-train set, 16 for meta-validation set, and 20 for meta-test set as in [28]. Similarly, tiered Image Net is composed of 608 classes...where 20 / 6 / 8 disjoint categories are used as meta-train / meta-validation / meta-test sets, respectively.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments, such as GPU models, CPU types, or memory specifications.
Software Dependencies No The paper does not provide specific version numbers for software components or libraries used (e.g., "Python 3.8", "PyTorch 1.9").
Experiment Setup Yes In the meta-training stage, the meta-learner gφ is trained over 100 epochs (each epoch with 500 iterations) with a batch size of 2 and 4 for 5-shot and 1-shot, respectively. At each iteration, we sample N classes for N-way classification, followed by sampling k labeled examples for Di and 15 examples for D i for each class.