search

CampanulaBells / PREM-GAD

Code for IEEE ICDM 23 PREM: A Simple Yet Effective Approach for Node-Level Graph Anomaly Detection
10 stars 1 forks source link

Datasets #2

Open Djerry-h opened 3 months ago

Djerry-h commented 3 months ago

Hello, it's an honor to read your article, but I have some questions. Did the model consider real-world data that contains anomalies (e.g., YelpChi) during the experiments? I wonder if any related experimental analysis has been done?

CampanulaBells commented 3 months ago

Hi, we haven't tested it on other datasets yet. You can have a try!

Djerry-h commented 3 months ago

Sure, Professor, I have another small question. In the document, it mentions the dataset attributes. What does that refer to? Because when I printed Flicker, I didn't find the 12,407-dimensional value mentioned in the dataset. Could you take a look at the image I provided? 1717417797670 1717418020318

CampanulaBells commented 3 months ago

Hi, attributes refer to feature dimension. The 7575 is the number of nodes. So basically it is a matrix of node's features.

Djerry-h commented 3 months ago

Thank you for your patient response！  

deom @.***

 

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2024年6月4日(星期二) 上午7:59 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [CampanulaBells/PREM-GAD] Datasets (Issue #2)

Hi, attributes refer to feature dimension. The 7575 is the number of nodes. So basically it is a matrix of node's features.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>