Meta-Learning by Adjusting Priors Based on Extended PAC-Bayes Theory

Authors: Ron Amit, Ron Meir

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section we demonstrate the performance of our transfer method with image classification tasks solved by deep neural networks. In image classification, the data samples, z (x, y), consist of a an image, x, and a label, y. The hypothesis class hw : w Rd is the set of neural networks with a given architecture (which will be specified later). As a loss function ℓ(hw, z) we will use the cross-entropy loss.
Researcher Affiliation Academia 1The Viterbi Faculty of Electrical Engineering, Technion Israel Institute of Technology, Haifa, Israel. Correspondence to: Ron Amit <ronamit@campus.technion.ac.il>, Ron Meir <rmeir@ee.technion.ac.il>.
Pseudocode Yes Both algorithms are described in pseudo-code in the supplementary material (section A.4) 11 12.
Open Source Code Yes 11Code is available at: https://github.com/ ron-amit/meta-learning-adjusting-priors.
Open Datasets Yes We conduct two experiments with two different task environments, based on augmentations of the MNIST dataset (Le Cun, 1998).
Dataset Splits No The paper mentions 'meta-training set' and 'meta-test task' sizes but does not specify a separate validation set or its split information.
Hardware Specification Yes We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan Xp GPU used for this research.
Software Dependencies No The paper mentions using deep neural networks but does not provide specific software dependencies or their version numbers (e.g., Python, TensorFlow, PyTorch versions).
Experiment Setup No The paper states 'See section A.5 for more implementation details,' indicating that specific experimental setup details like hyperparameters are not present in the main text.