PACIA: Parameter-Efficient Adapter for Few-Shot Molecular Property Prediction
Authors: Shiguang Wu, Yaqing Wang, Quanming Yao
IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive results show that PACIA obtains the stateof-the-art performance in few-shot MPP problems, and our proposed hierarchical adaptation mechanism is rational and effective.In this section, we evaluate the proposed PACIA2 on few-shot MPP problems. We run all experiments with 10 random seeds, and report the mean and standard deviations (in the subscript bracket). |
| Researcher Affiliation | Collaboration | Shiguang Wu1 , Yaqing Wang2 , Quanming Yao1 1Department of Electronic Engineering, Tsinghua University 2Baidu Research, Baidu Inc. wsg23@mails.tsinghua.edu.cn, wangyaqing01@baidu.com, qyaoaa@tsinghua.edu.cn |
| Pseudocode | Yes | Algorithm 1 Meta-training procedure of PACIA. |
| Open Source Code | Yes | Code is available at https://github.com/LARS-research/PACIA. |
| Open Datasets | Yes | We use Tox21 [National Center for Advancing Translational Sciences, 2017], SIDER [Kuhn et al., 2016], MUV [Rohrer and Baumann, 2009] and Tox Cast [Richard et al., 2016] from Molecule Net [Wu et al., 2018], which are commonly used to evaluate the performance on few-shot MPP [Altae-Tran et al., 2017; Wang et al., 2021].We also use FS-Mol [Stanley et al., 2021], a new benchmark consisting of a large number of diverse tasks for model pretraining and a set of few-shot tasks with imbalanced classes. |
| Dataset Splits | Yes | Following earlier works [Altae-Tran et al., 2017; Stanley et al., 2021; Chen et al., 2022; Schimunek et al., 2023], we model a Tτ as a 2-way classification task Tτ, associating with a support set Sτ = {(Xτ,s, yτ,s)}Nτ s=1 containing labeled samples from active/inactive class, and a query set Qτ = {(Xτ,q, yτ,q)}Mτ q=1 containing Mτ samples whose labels are only used for evaluation. We consider both (i) balanced support sets, i.e., Sτ contains Nτ 2 samples per class which is consistent with the standard N-way K-shot FSL setting [Altae-Tran et al., 2017], and (ii) imbalanced support sets which exist in real-world applications [Stanley et al., 2021].The support sets are balanced, each of them contains K labeled molecules per class, where K = 1 and K = 10 are considered.Each support set contains 64 labeled molecules, and can be imbalanced where the number of labeled molecules from active and inactive is not equal. All remaining molecules in the task form the query set. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper mentions using GIN as an encoder, but does not provide specific version numbers for any software dependencies like deep learning frameworks or libraries. |
| Experiment Setup | Yes | We run all experiments with 10 random seeds, and report the mean and standard deviations (in the subscript bracket).Following earlier works [Guo et al., 2021; Wang et al., 2021], we use GIN [Xu et al., 2019] as encoder, which is trained from scratch.The support sets are balanced, each of them contains K labeled molecules per class, where K = 1 and K = 10 are considered.We consider various variants of PACIA, including (i) finetuning: using the same model structure and fine-tuning all parameters to adapt to each property without hypernetworks; (ii) w/o T: removing task-level adaptation, thus the GNN encoder will not be adapted by hypernetworks w.r.t. each property; and (iii) w/o Q: removing query-level adaptation, such that all molecules are processed by the same predictor. |