repository
Evidential Meta-Learning for Molecular Property Prediction
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | 이슬 | - |
| dc.contributor.author | 함경표 | - |
| dc.date.accessioned | 2025-01-25T01:36:11Z | - |
| dc.date.available | 2025-01-25T01:36:11Z | - |
| dc.date.issued | 2023-08 | - |
| dc.identifier.other | 32992 | - |
| dc.identifier.uri | https://dspace.ajou.ac.kr/handle/2018.oak/24711 | - |
| dc.description | 학위논문(석사)--아주대학교 일반대학원 :인공지능학과,2023. 8 | - |
| dc.description.tableofcontents | 1 Introduction 1 <br>2 Method 4 <br> 2.1 Evidential Meta Learning Framework 5 <br> 2.1.1 EM3P2 Meta Training 6 <br> 2.1.2 EM3P2 Meta Testing 9 <br> 2.2 Graph Embedding and Classification Model 10 <br> 2.2.1 Graph Isomorphism Network 10 <br> 2.2.2 Evidential Multi-layer Perceptron(EMLP) 10 <br> 2.2.3 Estimating Uncertainty 11 <br> 2.3 Loss Function 12 <br> 2.3.1 Evidential Loss Function 12 <br> 2.3.2 Belief Regularizer 13 <br> 2.3.3 Accuracy Versus Uncertainty Loss Calibration 13 <br> 2.3.4 Overall Loss Function 14 <br>3 Experiment 15 <br> 3.1 Dataset 15 <br> 3.2 Data Summary 16 <br> 3.2.1 Tox21 16 <br> 3.2.2 Sider 17 <br> 3.3 Compared Methods 19 <br> 3.3.1 Reproducibility Setting 19 <br> 3.3.2 Evaluation 20 <br>4 Results 23 <br> 4.1 Comparison with State-of-the-art 23 <br> 4.2 Case Study of accuracy according to vacuity threshold 29 <br> 4.3 Case Study of Query balancing 31 <br> 4.4 Case Study of Belief Quantification with Softmax 32 <br>5 Conclusion 35 | - |
| dc.language.iso | eng | - |
| dc.publisher | The Graduate School, Ajou University | - |
| dc.rights | 아주대학교 논문은 저작권에 의해 보호받습니다. | - |
| dc.title | Evidential Meta-Learning for Molecular Property Prediction | - |
| dc.type | Thesis | - |
| dc.contributor.affiliation | 아주대학교 대학원 | - |
| dc.contributor.department | 일반대학원 인공지능학과 | - |
| dc.date.awarded | 2023-08 | - |
| dc.description.degree | Master | - |
| dc.identifier.localId | T000000032992 | - |
| dc.identifier.url | https://dcoll.ajou.ac.kr/dcollection/common/orgView/000000032992 | - |
| dc.subject.keyword | Evidential Neural Network | - |
| dc.subject.keyword | Few-shot Learning | - |
| dc.subject.keyword | Meta-Learning | - |
| dc.subject.keyword | Molecular Property Prediction | - |
| dc.description.alternativeAbstract | The usefulness of supervised molecular property prediction is well-recognized in many applications. However, inadequacy and imbalance of labeled data make the learning problem difficult. Moreover, the reliability of the predictions is also a huddle in the distribution of supervised molecular property prediction in the cost and safety-critical application fields, such as drug discovery. We propose an EM3P2 model as a supervised molecular property prediction method that addresses the problem of data insufficiency and reliability. Our proposed method trains an evidential graph isomorphism network classifier using multi-task molecular property datasets on top of a model-agnostic meta-learning (MAML) scheme. <br>Our model is a well-orchestrated combination of evidential neural networks for estimating model prediction uncertainty, graph isomorphism networks for embedding vector input molecular graphs, and data balance-aware model-agnostic meta-learning for generating a meta-model that adapts to new tasks with little labeled data. | - |
- Appears in Collections:
- Graduate School of Ajou University > Department of Artificial Intelligence > 3. Theses(Master)
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
-
- License
- STATISTICS
- Total Visit :5,012,209
- Total Download :2,095
- Today View :2,392
