UB-GOLD
收藏资源简介:
UB-GOLD数据集由吉林大学创建,包含35个数据集,覆盖四种实际的异常和分布外检测场景。数据集内容丰富,涉及生物信息学、社交网络等多个领域,旨在通过不同类型的异常和分布外样本,评估和比较16种代表性的GLAD/GLOD方法。创建过程中,数据集被精心设计以模拟不同的数据分布和异常类型。该数据集主要应用于图机器学习系统的安全性和可靠性评估,解决图级异常检测和分布外检测的问题。
The UB-GOLD dataset was created by Jilin University, consisting of 35 datasets covering four practical anomaly and out-of-distribution (OOD) detection scenarios. With rich content spanning multiple fields including bioinformatics and social networks, this dataset aims to evaluate and compare 16 representative GLAD/GLOD methods using diverse anomalous and out-of-distribution samples. During its development, the dataset was meticulously designed to simulate different data distributions and anomaly types. Primarily applied to the safety and reliability assessment of graph machine learning systems, this dataset addresses the problems of graph-level anomaly detection and out-of-distribution detection.
UB-GOLD 数据集概述
数据集来源
UB-GOLD 数据集包含以下几种类型的数据:
- 内在异常:TOX21
- Tox21_p53, Tox21_HSE, Tox21_MMP, Tox21_PPAR-gamma
- 跨数据集偏移与基于类别的异常:TUDataset
- COLLAB, IMDB-BINARY, REDDIT-BINARY, ENZYMES, PROTEINS
- DD, BZR, AIDS, COX2, NCI1, DHFR
- 跨数据集偏移:OGB
- BBBP, BACE, CLINTOX, LIPO, FREESOLV
- TOXCAST, SOL, MUV, TOX21, SIDER
- 数据集内偏移:DrugOOD
- IC50 (SIZE, SCAFFOLD, ASSAY)
- EC50 (SIZE, SCAFFOLD, ASSAY)
- 数据集内偏移:GOOD
支持的方法
UB-GOLD 支持以下16种流行的异常检测和OOD检测方法:
表1:2-Step方法
表2:端到端方法




