pith. sign in

arxiv: 2402.10380 · v1 · pith:LNKQRBVYnew · submitted 2024-02-16 · 💻 cs.LG · cs.AI· cs.CL

Subgraph-level Universal Prompt Tuning

Junhyun Lee , Wooseong Yang , Jaewoo Kang This is my paper

Pith reviewed 2026-05-24 03:35 UTC · model grok-4.3

classification 💻 cs.LG cs.AIcs.CL
keywords prompt tuninggraph neural networksuniversal promptsubgraph-level tuningfew-shot learningpre-trained graph models
0
0 comments X

The pith

Assigning prompt features at the subgraph level lets a universal prompt tuning method capture complex graph contexts while using far fewer parameters than fine-tuning.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces Subgraph-level Universal Prompt Tuning to address the limitation of simple feature-space prompts in understanding complex graph structures. By assigning prompts at the subgraph level, SUPT maintains its ability to work with any pre-training strategy. This approach requires significantly fewer tuning parameters compared to fine-tuning methods. It demonstrates superior performance in most experiments on graph tasks under both full-shot and few-shot scenarios.

Core claim

In SUPT, prompt features are assigned at the subgraph-level, preserving the method's universal capability. This requires extremely fewer tuning parameters than fine-tuning-based methods.

What carries the argument

Subgraph-level assignment of prompt features, which allows capturing detailed local contexts in graphs while emulating any prompting function in the feature space.

If this is right

  • Outperforms fine-tuning-based methods in 42 out of 45 full-shot scenario experiments with an average improvement of over 2.5%.
  • Excels in 41 out of 45 few-shot experiments with an average performance increase of more than 6.6%.
  • Applies across different pre-training strategies for graph neural networks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If subgraph prompts succeed by focusing on local structure, similar granularity adjustments might improve prompting in other data types like sequences.
  • The efficiency gain suggests testing whether even coarser or adaptive levels of prompt assignment could further reduce parameters on very large graphs.

Load-bearing premise

That assigning prompts at the subgraph level will allow the method to grasp the complex contexts in graphs that simpler feature-space prompts could not.

What would settle it

A set of experiments on graphs with intricate local structures where subgraph-level prompts show no performance gain over feature-space prompts would disprove the claimed improvement.

Figures

Figures reproduced from arXiv: 2402.10380 by Jaewoo Kang, Junhyun Lee, Wooseong Yang.

Figure 1
Figure 1. Figure 1: Illustration of the concepts: (a) pixel-level visual [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of Different Tuning Approaches: Comparing Fine-Tuning, GPF, GPF-Plus, and SUPT. GPF assigns a [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
read the original abstract

In the evolving landscape of machine learning, the adaptation of pre-trained models through prompt tuning has become increasingly prominent. This trend is particularly observable in the graph domain, where diverse pre-training strategies present unique challenges in developing effective prompt-based tuning methods for graph neural networks. Previous approaches have been limited, focusing on specialized prompting functions tailored to models with edge prediction pre-training tasks. These methods, however, suffer from a lack of generalizability across different pre-training strategies. Recently, a simple prompt tuning method has been designed for any pre-training strategy, functioning within the input graph's feature space. This allows it to theoretically emulate any type of prompting function, thereby significantly increasing its versatility for a range of downstream applications. Nevertheless, the capacity of such simple prompts to fully grasp the complex contexts found in graphs remains an open question, necessitating further investigation. Addressing this challenge, our work introduces the Subgraph-level Universal Prompt Tuning (SUPT) approach, focusing on the detailed context within subgraphs. In SUPT, prompt features are assigned at the subgraph-level, preserving the method's universal capability. This requires extremely fewer tuning parameters than fine-tuning-based methods, outperforming them in 42 out of 45 full-shot scenario experiments with an average improvement of over 2.5%. In few-shot scenarios, it excels in 41 out of 45 experiments, achieving an average performance increase of more than 6.6%.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces Subgraph-level Universal Prompt Tuning (SUPT) for graph neural networks. Building on prior simple universal prompts that operate in feature space and can emulate any prompting function, SUPT assigns prompt features at the subgraph level to address the open question of capturing complex graph contexts. The method is claimed to remain universal across pre-training strategies, require far fewer parameters than fine-tuning, and deliver empirical gains: outperforming fine-tuning baselines in 42/45 full-shot experiments (average improvement >2.5%) and 41/45 few-shot experiments (average >6.6%).

Significance. If the experimental results prove reproducible with proper controls, the work could meaningfully advance general-purpose prompt tuning for GNNs by providing a subgraph-aware yet still universal and parameter-efficient alternative to model-specific or full fine-tuning approaches.

major comments (2)
  1. [Abstract] Abstract: the central empirical claims (42/45 full-shot wins with >2.5% average gain; 41/45 few-shot wins with >6.6% average gain) are stated without any description of experimental design, datasets, pre-training tasks, subgraph assignment implementation, number of runs, variance, or statistical tests. This information is load-bearing for assessing whether the data support the outperformance claim.
  2. [Abstract] Abstract: the assertion that subgraph-level assignment enables the method to 'grasp the complex contexts found in graphs' (the identified open question) is presented without any accompanying mechanism, ablation, or concrete comparison showing that subgraph granularity, rather than other factors, is responsible for the reported gains.
minor comments (1)
  1. [Abstract] The sentence 'This requires extremely fewer tuning parameters' is grammatically awkward; rephrase for clarity (e.g., 'This uses far fewer tuning parameters').

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below, clarifying that detailed experimental information and mechanisms appear in the body of the manuscript while proposing targeted abstract revisions where they improve clarity without exceeding length constraints.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central empirical claims (42/45 full-shot wins with >2.5% average gain; 41/45 few-shot wins with >6.6% average gain) are stated without any description of experimental design, datasets, pre-training tasks, subgraph assignment implementation, number of runs, variance, or statistical tests. This information is load-bearing for assessing whether the data support the outperformance claim.

    Authors: We agree the abstract is concise and omits these specifics. Section 4 fully specifies the experimental design (9 datasets, 5 pre-training strategies, subgraph assignment via the procedure in Section 3.2, results as mean ± std over 10 independent runs, and paired t-tests for significance). To address the concern, we will add one sentence to the abstract summarizing the experimental scope while preserving brevity. revision: partial

  2. Referee: [Abstract] Abstract: the assertion that subgraph-level assignment enables the method to 'grasp the complex contexts found in graphs' (the identified open question) is presented without any accompanying mechanism, ablation, or concrete comparison showing that subgraph granularity, rather than other factors, is responsible for the reported gains.

    Authors: The mechanism is presented in Section 3: prompt features are assigned per subgraph (extracted via the described partitioning) to encode local contexts while the universal property is retained by operating in feature space. Section 5.3 reports ablations (subgraph-level vs. node-level vs. graph-level) that isolate the granularity contribution to the observed gains. The abstract summarizes this contribution; the supporting evidence resides in the main text. revision: no

Circularity Check

0 steps flagged

No significant circularity; empirical method with external validation

full rationale

The paper presents SUPT as a subgraph-level prompt tuning technique for GNNs, claiming it preserves universality while using fewer parameters and outperforming baselines in 42/45 full-shot and 41/45 few-shot experiments. No equations, derivations, or load-bearing self-citations appear in the provided abstract or described content. The central claims rest on reported experimental margins rather than any mathematical reduction to author-defined inputs or prior self-work. This is the most common honest finding for an empirical methods paper whose results are benchmarked externally.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the empirical superiority of subgraph-level prompts; the key unverified premise is that subgraph granularity solves the context-capture limitation of simpler prompts. No free parameters, axioms, or invented entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption Subgraph-level prompt assignment can capture complex graph contexts that node- or graph-level prompts cannot while remaining universal across pre-training strategies.
    This premise is invoked to justify moving from the simple universal prompt to SUPT.

pith-pipeline@v0.9.0 · 5783 in / 1351 out tokens · 47660 ms · 2026-05-24T03:35:27.269176+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

67 extracted references · 67 canonical work pages · 3 internal anchors

  1. [1]

    Hyojin Bahng, Ali Jahanian, Swami Sankaranarayanan, and Phillip Isola. 2022. Exploring Visual Prompts for Adapting Large-Scale Models. arXiv preprint arXiv:2203.17274 (2022)

    work page arXiv 2022

  2. [2]

    Filippo Maria Bianchi, Daniele Grattarola, and Cesare Alippi. 2020. Spectral clus- tering with graph neural networks for graph pooling. In International conference on machine learning. PMLR, 874–883

    work page 2020

  3. [3]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901

    work page 2020

  4. [4]

    Aochuan Chen, Peter Lorenz, Yuguang Yao, Pin-Yu Chen, and Sijia Liu. 2023. Visual Prompting for Adversarial Robustness. In ICASSP 2023 - 2023 IEEE In- ternational Conference on Acoustics, Speech and Signal Processing (ICASSP) . 1–5. https://doi.org/10.1109/ICASSP49357.2023.10097245

    work page doi:10.1109/icassp49357.2023.10097245 2023

  5. [5]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. arXiv preprint arXiv:1810.04805 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  6. [6]

    Yifan Du, Zikang Liu, Junyi Li, and Wayne Xin Zhao. 2022. A Survey of Vision- Language Pre-Trained Models. InProceedings of the Thirty-First International Joint Conference on Artificial Intelligence, IJCAI-22, Lud De Raedt (Ed.). International Joint Conferences on Artificial Intelligence Organization, 5436–5443. https: //doi.org/10.24963/ijcai.2022/762 Su...

    work page doi:10.24963/ijcai.2022/762 2022

  7. [7]

    Taoran Fang, Yunchao Mercer Zhang, Yang Yang, Chunping Wang, and Lei CHEN

    work page

  8. [8]

    In Thirty-seventh Conference on Neural Information Processing Systems

    Universal Prompt Tuning for Graph Neural Networks. In Thirty-seventh Conference on Neural Information Processing Systems . https://openreview.net/ forum?id=0LmWBhIYLi

    work page

  9. [9]

    Hongyang Gao and Shuiwang Ji. 2019. Graph u-nets. In international conference on machine learning. PMLR, 2083–2092

    work page 2019

  10. [10]

    Anna Gaulton, Louisa J Bellis, A Patricia Bento, Jon Chambers, Mark Davies, Anne Hersey, Yvonne Light, Shaun McGlinchey, David Michalovich, Bissan Al-Lazikani, et al. 2012. ChEMBL: a large-scale bioactivity database for drug discovery.Nucleic acids research 40, D1 (2012), D1100–D1107

    work page 2012

  11. [11]

    Xueting Han, Zhenhuan Huang, Bang An, and Jing Bai. 2021. Adaptive transfer learning on graph neural networks. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining . 565–574

    work page 2021

  12. [12]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition . 770–778

    work page 2016

  13. [13]

    Weihua Hu, Bowen Liu, Joseph Gomes, Marinka Zitnik, Percy Liang, Vijay Pande, and Jure Leskovec. 2020. Strategies for Pre-training Graph Neural Networks. In International Conference on Learning Representations . https://openreview.net/ forum?id=HJlWWJSFDH

    work page 2020

  14. [14]

    Ziniu Hu, Yuxiao Dong, Kuansan Wang, Kai-Wei Chang, and Yizhou Sun. 2020. Gpt-gnn: Generative pre-training of graph neural networks. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 1857–1867

    work page 2020

  15. [15]

    Qian Huang, Hongyu Ren, Peng Chen, Gregor Kržmanc, Daniel Zeng, Percy Liang, and Jure Leskovec. 2023. PRODIGY: Enabling In-context Learning Over Graphs. In Thirty-seventh Conference on Neural Information Processing Systems . https://openreview.net/forum?id=pLwYhNNnoR

    work page 2023

  16. [16]

    Amir Hosein Khasahmadi, Kaveh Hassani, Parsa Moradi, Leo Lee, and Quaid Morris. 2020. Memory-Based Graph Networks. In International Conference on Learning Representations. https://openreview.net/forum?id=r1laNeBYPB

    work page 2020

  17. [17]

    Adam: A Method for Stochastic Optimization

    Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Opti- mization. In 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings , Yoshua Bengio and Yann LeCun (Eds.). http://arxiv.org/abs/1412.6980

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  18. [18]

    Thomas N Kipf and Max Welling. 2016. Variational Graph Auto-Encoders. NIPS Workshop on Bayesian Deep Learning (2016)

    work page 2016

  19. [19]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In International Conference on Learning Repre- sentations. https://openreview.net/forum?id=SJU4ayYgl

    work page 2017

  20. [20]

    Junhyun Lee, Inyeop Lee, and Jaewoo Kang. 2019. Self-attention graph pooling. In International conference on machine learning . PMLR, 3734–3743

    work page 2019

  21. [21]

    Brian Lester, Rami Al-Rfou, and Noah Constant. 2021. The Power of Scale for Parameter-Efficient Prompt Tuning. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing , Marie-Francine Moens, Xuanjing Huang, Lucia Specia, and Scott Wen-tau Yih (Eds.). Association for Computational Linguistics, Online and Punta Cana, Domini...

    work page 2021

  22. [22]

    https://doi.org/10.18653/v1/2021.emnlp-main.243

    work page doi:10.18653/v1/2021.emnlp-main.243 2021

  23. [23]

    Guohao Li, Matthias Müller, Bernard Ghanem, and Vladlen Koltun. 2021. Training graph neural networks with 1000 layers. In International conference on machine learning. PMLR, 6437–6449

    work page 2021

  24. [24]

    Xiang Lisa Li and Percy Liang. 2021. Prefix-Tuning: Optimizing Continuous Prompts for Generation. In Proceedings of the 59th Annual Meeting of the Associa- tion for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , Chengqing Zong, Fei Xia, Wenjie Li, and Roberto Navigli (Eds.). A...

    work page doi:10.18653/v1/2021.acl-long.353 2021

  25. [25]

    Yizhe Li, Yu-Lin Tsai, Chia-Mu Yu, Pin-Yu Chen, and Xuebin Ren. 2023. Exploring the benefits of visual prompting in differential privacy. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 5158–5167

    work page 2023

  26. [26]

    Huihui Liu, Yiding Yang, and Xinchao Wang. 2021. Overcoming catastrophic forgetting in graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, Vol. 35. 8653–8661

    work page 2021

  27. [27]

    Meng Liu, Hongyang Gao, and Shuiwang Ji. 2020. Towards deeper graph neural networks. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining . 338–348

    work page 2020

  28. [28]

    Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neubig. 2023. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. Comput. Surveys 55, 9 (2023), 1–35

    work page 2023

  29. [29]

    Weihuang Liu, Xi Shen, Chi-Man Pun, and Xiaodong Cun. 2023. Explicit visual prompting for low-level structure segmentations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 19434–19445

    work page 2023

  30. [30]

    Xiao Liu, Yanan Zheng, Zhengxiao Du, Ming Ding, Yujie Qian, Zhilin Yang, and Jie Tang. 2023. GPT understands, too. AI Open (2023)

    work page 2023

  31. [31]

    Zemin Liu, Xingtong Yu, Yuan Fang, and Xinming Zhang. 2023. Graphprompt: Unifying pre-training and downstream tasks for graph neural networks. In Pro- ceedings of the ACM Web Conference 2023 . 417–428

    work page 2023

  32. [32]

    Yao Ma, Suhang Wang, Charu C Aggarwal, and Jiliang Tang. 2019. Graph convo- lutional networks with eigenpooling. In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining . 723–731

    work page 2019

  33. [33]

    Zheng Ma, Junyu Xuan, Yu Guang Wang, Ming Li, and Pietro Liò. 2020. Path integral based convolution and pooling for graph neural networks. Advances in Neural Information Processing Systems 33 (2020), 16421–16433

    work page 2020

  34. [34]

    Andreas Mayr, Günter Klambauer, Thomas Unterthiner, Marvin Steijaert, Jörg K Wegner, Hugo Ceulemans, Djork-Arné Clevert, and Sepp Hochreiter. 2018. Large- scale comparison of machine learning methods for drug target prediction on ChEMBL. Chemical science 9, 24 (2018), 5441–5451

    work page 2018

  35. [35]

    Hoang Nt and Takanori Maehara. 2019. Revisiting graph neural networks: All we have is low-pass filters. arXiv preprint arXiv:1905.09550 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  36. [36]

    Changdae Oh, Hyeji Hwang, Hee-young Lee, YongTaek Lim, Geunyoung Jung, Jiyoung Jung, Hosik Choi, and Kyungwoo Song. 2023. BlackVIP: Black-Box Visual Prompting for Robust Transfer Learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . 24224–24235

    work page 2023

  37. [37]

    Kenta Oono and Taiji Suzuki. 2020. Graph Neural Networks Exponentially Lose Expressive Power for Node Classification. InInternational Conference on Learning Representations. https://openreview.net/forum?id=S1ldO2EFPr

    work page 2020

  38. [38]

    Jiezhong Qiu, Qibin Chen, Yuxiao Dong, Jing Zhang, Hongxia Yang, Ming Ding, Kuansan Wang, and Jie Tang. 2020. Gcc: Graph contrastive coding for graph neural network pre-training. InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining . 1150–1160

    work page 2020

  39. [39]

    Xipeng Qiu, Tianxiang Sun, Yige Xu, Yunfan Shao, Ning Dai, and Xuanjing Huang

    work page

  40. [40]

    Science China Technological Sciences 63, 10 (2020), 1872–1897

    Pre-trained models for natural language processing: A survey. Science China Technological Sciences 63, 10 (2020), 1872–1897

    work page 2020

  41. [41]

    Teague Sterling and John J Irwin. 2015. ZINC 15–ligand discovery for everyone. Journal of chemical information and modeling 55, 11 (2015), 2324–2337

    work page 2015

  42. [42]

    Fan-Yun Sun, Jordan Hoffman, Vikas Verma, and Jian Tang. 2019. InfoGraph: Unsupervised and Semi-supervised Graph-Level Representation Learning via Mutual Information Maximization. In International Conference on Learning Rep- resentations

    work page 2019

  43. [43]

    Mingchen Sun, Kaixiong Zhou, Xin He, Ying Wang, and Xin Wang. 2022. Gppt: Graph pre-training and prompt tuning to generalize graph neural networks. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 1717–1727

    work page 2022

  44. [44]

    Xiangguo Sun, Hong Cheng, Jia Li, Bo Liu, and Jihong Guan. 2023. All in One: Multi-Task Prompting for Graph Neural Networks. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (<conf-loc>, <city>Long Beach</city>, <state>CA</state>, <country>USA</country>, </conf- loc>) (KDD ’23). Association for Computing Machinery, N...

    work page doi:10.1145/3580305.3599256 2023

  45. [45]

    Susheel Suresh, Pan Li, Cong Hao, and Jennifer Neville. 2021. Adversarial graph augmentation to improve graph contrastive learning. Advances in Neural Infor- mation Processing Systems 34 (2021), 15920–15933

    work page 2021

  46. [46]

    Mingxing Tan and Quoc Le. 2019. Efficientnet: Rethinking model scaling for convolutional neural networks. In International conference on machine learning . PMLR, 6105–6114

    work page 2019

  47. [47]

    Hsi-Ai Tsao, Lei Hsiung, Pin-Yu Chen, Sijia Liu, and Tsung-Yi Ho. 2024. AutoVP: An Automated Visual Prompting Framework and Benchmark. In The Twelfth International Conference on Learning Representations

    work page 2024

  48. [48]

    Hamilton, Pietro Liò, Yoshua Bengio, and R Devon Hjelm

    Petar Veličković, William Fedus, William L. Hamilton, Pietro Liò, Yoshua Bengio, and R Devon Hjelm. 2019. Deep Graph Infomax. In International Conference on Learning Representations. https://openreview.net/forum?id=rklz9iAcKQ Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Lee et al

    work page 2019

  49. [49]

    Felix Wu, Amauri Souza, Tianyi Zhang, Christopher Fifty, Tao Yu, and Kilian Weinberger. 2019. Simplifying graph convolutional networks. In International conference on machine learning . PMLR, 6861–6871

    work page 2019

  50. [50]

    Xinyi Wu, Amir Ajorlou, Zihui Wu, and Ali Jadbabaie. 2023. Demystifying Oversmoothing in Attention-Based Graph Neural Networks. In Thirty-seventh Conference on Neural Information Processing Systems . https://openreview.net/ forum?id=Kg65qieiuB

    work page 2023

  51. [51]

    Zhenqin Wu, Bharath Ramsundar, Evan N Feinberg, Joseph Gomes, Caleb Ge- niesse, Aneesh S Pappu, Karl Leswing, and Vijay Pande. 2018. MoleculeNet: a benchmark for molecular machine learning.Chemical science 9, 2 (2018), 513–530

    work page 2018

  52. [52]

    Jun Xia, Lirong Wu, Jintao Chen, Bozhen Hu, and Stan Z Li. 2022. Simgrace: A simple framework for graph contrastive learning without data augmentation. In Proceedings of the ACM Web Conference 2022 . 1070–1079

    work page 2022

  53. [53]

    Jun Xia, Jiangbin Zheng, Cheng Tan, Ge Wang, and Stan Z Li. 2022. Towards effective and generalizable fine-tuning for pre-trained molecular graph models. bioRxiv (2022), 2022–02

    work page 2022

  54. [54]

    Jun Xia, Yanqiao Zhu, Yuanqi Du, and Stan Z Li. 2022. A survey of pretraining on graphs: Taxonomy, methods, and applications. arXiv preprint arXiv:2202.07893 (2022)

    work page arXiv 2022

  55. [55]

    Yinghui Xing, Qirui Wu, De Cheng, Shizhou Zhang, Guoqiang Liang, Peng Wang, and Yanning Zhang. 2023. Dual Modality Prompt Tuning for Vision- Language Pre-Trained Model. IEEE Transactions on Multimedia (2023), 1–13. https://doi.org/10.1109/TMM.2023.3291588

    work page doi:10.1109/tmm.2023.3291588 2023

  56. [56]

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. 2019. How Powerful are Graph Neural Networks?. In International Conference on Learning Representa- tions. https://openreview.net/forum?id=ryGs6iA5Km

    work page 2019

  57. [57]

    Keyulu Xu, Chengtao Li, Yonglong Tian, Tomohiro Sonobe, Ken-ichi Kawarabayashi, and Stefanie Jegelka. 2018. Representation learning on graphs with jumping knowledge networks. In International conference on machine learn- ing. PMLR, 5453–5462

    work page 2018

  58. [58]

    Minghao Xu, Hang Wang, Bingbing Ni, Hongyu Guo, and Jian Tang. 2021. Self- supervised graph-level representation learning with local and global structure. In International Conference on Machine Learning . PMLR, 11548–11558

    work page 2021

  59. [59]

    Zhitao Ying, Jiaxuan You, Christopher Morris, Xiang Ren, Will Hamilton, and Jure Leskovec. 2018. Hierarchical graph representation learning with differentiable pooling. Advances in neural information processing systems 31 (2018)

    work page 2018

  60. [60]

    Yuning You, Tianlong Chen, Yongduo Sui, Ting Chen, Zhangyang Wang, and Yang Shen. 2020. Graph contrastive learning with augmentations. Advances in neural information processing systems 33 (2020), 5812–5823

    work page 2020

  61. [61]

    Yuning You, Tianlong Chen, Zhangyang Wang, and Yang Shen. 2022. Bringing your own view: Graph contrastive learning without prefabricated data augmenta- tions. In Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining. 1300–1309

    work page 2022

  62. [62]

    Hao Yuan and Shuiwang Ji. 2020. StructPool: Structured Graph Pooling via Con- ditional Random Fields. In International Conference on Learning Representations . https://openreview.net/forum?id=BJxg_hVtwH

    work page 2020

  63. [63]

    Liang Zhang, Xudong Wang, Hongsheng Li, Guangming Zhu, Peiyi Shen, Ping Li, Xiaoyuan Lu, Syed Afaq Ali Shah, and Mohammed Bennamoun. 2020. Structure- feature based graph self-adaptive pooling. In Proceedings of The Web Conference

    work page 2020

  64. [64]

    Zaixi Zhang, Qi Liu, Hao Wang, Chengqiang Lu, and Chee-Kong Lee. 2021. Motif- based graph self-supervised learning for molecular property prediction.Advances in Neural Information Processing Systems 34 (2021), 15870–15882

    work page 2021

  65. [65]

    Fan Zhou and Chengtai Cao. 2021. Overcoming catastrophic forgetting in graph neural networks with experience replay. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 4714–4722

    work page 2021

  66. [66]

    Kuangqi Zhou, Yanfei Dong, Kaixin Wang, Wee Sun Lee, Bryan Hooi, Huan Xu, and Jiashi Feng. 2021. Understanding and resolving performance degradation in deep graph convolutional networks. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management . 2728–2737

    work page 2021

  67. [67]

    Marinka Zitnik, Rok Sosič, Marcus W Feldman, and Jure Leskovec. 2019. Evolution of resilience in protein interactomes across the tree of life. Proceedings of the National Academy of Sciences 116, 10 (2019), 4426–4433. A PROOF FOR THEOREM 1 Theorem 1 (Universal Capability of SUPT ). Let 𝑓 be a pre- trained GNN model and G : (𝑋 , 𝐴) be an input graph with s...

    work page 2019