MIDAS
AI
graph anomaly detection
SHOW HN2020
How HN took it
5 positive, 6 neutral or asking, 0 mixed, 0 critical, out of 11 comments about the project. Thread score +0.46.
Idea: 2 for, 0 againstPerformance: 2 for, 0 againstDocs: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Impressive speed and results over prior methods
From 2 comments
Impressive results! Few questions though:- 1) Do you use sliding window approach or exponential decay? 2) When you say groups of similar edges, do they have to be spatially close…
shivin9 on HNGreat to see that! The results look really impressive!
raghavjajodia on HN
Real-time detection with theoretical bounds is valuable
From 2 comments
Before I read your paper, I didn't realize the power of hypothesis tests. I was surprised to see theoretical bounds on anomaly detection with such weak constraint through simple…
ZhuoCai on HNI found the paper to be quite exciting, and it is also solving one of the main tasks in anomaly detection, that is real-time anomaly detection. With other machine/deep learning…
riteshkumar99 on HN
Paper is clear and easy to follow
From 1 comment
Interesting work and great paper as anomaly detection in real time is very important now a days. Easy to understand as well.
ramarohit on HN
Objections
Thread has suspected astroturfing from new accounts
From 1 comment
The comments in this thread exhibit some serious astroturfing. Also the 6 unique commenters accounts all created in the last 12 days. Just sayin'
ugh123 on HN
Requests
Handle DDoS-like attacks spread across destinations and multiple servers
From 2 comments
Thanks. It would be nice to have similarity methods that are not dependent on spatial locality to detect DDoS like attacks.
shivin9 on HNAh alright, that makes sense. But it works only when the destination is the same. In a setting where there are multiple web-hosting servers, you would need to treat a group of…
shivin9 on HN
Extend to non-graph data such as video streaming
From 1 comment
Hmm... Interesting work! Can we extend this to find anomalies in other kinds of data like video streaming?
sinhayash on HN
Questions
Can attackers evade detection by timing attacks
From 1 comment
Impressive results! Few questions though:- 1) Do you use sliding window approach or exponential decay? 2) When you say groups of similar edges, do they have to be spatially close…
shivin9 on HN
Difference between microcluster detection and subgraph detection
From 1 comment
Great work! However, from what I understand, microcluster detection seems to be very similar to subgraph detection. Is there a difference between the two problems? And if yes,…
arjitjain on HN
Alternatives named
Subgraph detection methods
From 1 comment
Thanks! There is a subtle difference. We define microclusters within the category of anomalous edge detection as 'suddenly arriving groups of suspiciously similar edges' e.g.…
siddharthbhatia on HN