This August, welcome to DSSG's first-ever Technical Reading Group! We will deeply dive into MIDAS, an algorithm that made waves at AAAI this year with the primary author of the research, Siddharth Bhatia.
7:00pm - 7.05pm: Introducing DSSG's Technical Reading Group Series
7:05pm - 7.35pm: Deep Dive into the Paper (https://arxiv.org/abs/1911.04464)
7.35pm - end: Q&A
The paper: MIDAS (Microcluster-Based Detector of Anomalies in Edge Streams) uses unsupervised learning to detect anomalies in a streaming manner in real-time and has become a new baseline. It was designed keeping in mind the way recent sophisticated attacks occur. MIDAS can be used to detect intrusions, Denial of Service (DoS), Distributed Denial of Service (DDoS) attacks, financial fraud and fake ratings. It combines a chi-squared goodness-of-fit test with the Count-Min-Sketch (CMS) streaming data structures to get an anomaly score for each edge. It then incorporates temporal and spatial relations to achieve better performance. MIDAS provides theoretical guarantees on the false positives and is three orders of magnitude faster than existing state of the art solutions. Check it out at https://github.com/Stream-AD/MIDAS
This will be a space for both education and lively debate, so come with your thoughts on the paper. Here are some questions to get you started:
1. Do you assume that all attacks are happening at one server/location?
2. Can the anomaly detection be parallelised?
3. Since MIDAS is so fast, do you plan to apply it directly over the network/hardware?
4. What do you do for malicious edges? Do they also get stored?
Siddharth Bhatia is a PhD student at National University of Singapore. Siddharth's research is supported by a President's Graduate Fellowship and he has been recognized as a Young Researcher by the ACM Heidelberg Laureate Forum. Siddharth has done breakthrough research in streaming anomaly detection. He is also applying his research at Amazon Web Services (AWS) during the summer. For more details, please visit https://www.comp.nus.edu.sg/~sbhatia/
Hi everyone. Yes, we've heard your complaints about people rsvping and not showing up, and how there's space at the venue but no more rsvp slots. To resolve this, we seek your help with taking attendence at the events:
- For members who RSVP and no-show, they will receive -1 point. (RSVPs close 48 hrs before the event--please cancel your rsvp before this)
- For members who RSVP and show, they will receive +1 point.
- E.g., Bill RSVPs for 3 meetups. He shows up for only 1. Thus, he gets (-1) + (-1) + (+1) = -1 point.
- Members who have < -3 points will be banned.
- Members who have > 10 points will get VIP and can attend meetups without RSVP for a year
- Please approach our committee members to take your attendance. Thank you!