Cost-aware Graph Generation: A Deep Bayesian Optimization Approach

Authors: Jiaxu Cui, Bo Yang, Bingyi Sun, Jiming Liu7142-7150

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Intensive experiments conducted on two challenging real-world applications, including molecular discovery and neural architecture search, demonstrate its effectiveness and applicability. The results show that it can generate the optimal graphs and reduce the evaluation costs significantly compared to the state-of-the-art.
Researcher Affiliation Academia Jiaxu Cui,1,2 Bo Yang,1,2 Bingyi Sun,1,2,4 Jiming Liu 3 1College of Computer Science and Technology, Jilin University, China 2Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Jilin University, China 3Department of Computer Science, Hong Kong Baptist University, Hong Kong 4National Laboratory for Parallel and Distributed Processing, National University of Defense Technology, China
Pseudocode Yes Algorithm 1 Procedure of the CAGG
Open Source Code Yes Code available at: https://github.com/csjtx1021/CAGG
Open Datasets Yes QM9 (Ramakrishnan et al. 2014) is a benchmark dataset in quantum chemistry, which contains about 134K molecules with at most 9 heavy atoms. NASBench201 (Dong and Yang 2020) is a unified benchmark for most up-to-date cell-based NAS methods. Indoor (Torres-Sospedra et al. 2014) collects about 20K localization records of mobile devices. Slice (Graf et al. 2011) contains 53,500 CT images from 74 patients.
Dataset Splits No The paper mentions datasets used but does not explicitly state the specific training, validation, or test dataset splits (e.g., percentages or exact counts) for its experiments. While NASBench201 is a benchmark with predefined splits, the paper does not specify how they used these or other datasets for their own partitioning.
Hardware Specification Yes Google Cloud Platform cost is calculated based on the price of on-demand c2-standard-8 instances.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies, such as programming languages, libraries, or frameworks used for implementation.
Experiment Setup No The paper mentions overall budgets for experiments (e.g., 'maximum # Eval to 3,000', '12-hour budget') and refers to supplementary material for constraint details. However, it does not provide specific hyperparameters like learning rates, batch sizes, optimizers, or detailed training configurations in the main text.