Posterior-Guided Neural Architecture Search

Authors: Yizhou Zhou, Xiaoyan Sun, Chong Luo, Zheng-Jun Zha, Wenjun Zeng6973-6980

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

Reproducibility Variable Result LLM Response
Research Type Experimental We validate our PGNAS method on the fundamental image classification task. Results on Cifar-10, Cifar-100 and Image Net show that PGNAS achieves a good trade-off between precision and speed of search among NAS methods. For example, it takes 11 GPU days to search a very competitive architecture with 1.98% and 14.28% test errors on Cifar10 and Cifar100, respectively.
Researcher Affiliation Collaboration 1University of Science Technology of China, 2Microsoft Research Asia
Pseudocode Yes Algorithm 1: PGNAS
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository.
Open Datasets Yes We validate our PGNAS method on the fundamental image classification task. Results on Cifar-10, Cifar-100 and Image Net show that PGNAS achieves a good trade-off between precision and speed of search among NAS methods.
Dataset Splits Yes Then the sampled candidates are evaluated and ranked on a held-out validation dataset.
Hardware Specification Yes Our PGNAS is trained in an end-to-end way with the Stochastic Gradient Descent (SGD) using a single P40 GPU card for Cifar-10/Cifar-100 and 4 M40 GPU cards for Image Net.
Software Dependencies No The paper mentions using Stochastic Gradient Descent (SGD) as an optimization method but does not provide specific version numbers for any software libraries, frameworks (e.g., TensorFlow, PyTorch), or programming languages used.
Experiment Setup Yes Our PGNAS is trained in an end-to-end way with the Stochastic Gradient Descent (SGD) ... For every super-network, we insert a dropout layer after each convolution layer according to Eq. 16 to facilitate the computation of Eq. 15. ... Please refer to the supplementary material for more details of the supernetworks and all hyper-parameter settings used in this paper. ... Table. 5 shows the impact of temperature value τ. ... Table. 6. The performance improves along with the increase of number of samples.