MLHat: The First International Workshop on Deployable Machine Learning for Security Defense

August 24, 2020 — Virtual Space
co-located with 25TH ACM SIGKDD Conference on Knowledge Discovery and Data Mining


David Wagner (UC Berkeley)
Machine Learning for Security and Security for Machine Learning

I will examine several applications of machine learning to detecting malicious behavior, and discuss several challenges in this space, including the base rate problem, lack of training data, and occurrence of anomalous but benign events. I will discuss possible directions for tackling these problems and highlight a few open problems in the field. One special risk of using machine learning to detect malicious behavior is that bad actors might modify their behavior to fool the classifier, by exploiting so-called adversarial examples. I will outline several directions in the literature to harden machine learning against adversarial examples.

David Wagner is Professor of Computer Science at the University of California at Berkeley, working in the area of computer security. His research has analyzed and contributed to the security of cellular networks, 802.11 wireless networks, electronic voting systems, and other widely deployed systems. He currently serves as a member of the federal committee responsible for drafting technical standards for electronic voting systems.


MLHat 2020 will be conducted in virtual space. The meeting will have both pre-recorded videos as well as live broadcast. The pre-recorded videos contain detailed presentations for each papers and they will be accessible continuously during the workshop day via KDD 2020 virtual platform, the workshop website, and YouTube channel. We will conduct the live broadcast via Zoom.

The schedule for the live broadcast is as follow (time is in PST):

TimeScheduled Events
8:00am – 8:05amOpening Remark
8:05am – 9:15amKeynote (David Wagner)
Machine Learning for Security and Security for Machine Learning
9:15am – 09:35amTalks + Q&A I: Understanding the Adversaries (Session Chair: Ali Ahmadzadeh)
A Large-Scale Analysis of Attacker Activity in Compromised Enterprise Accounts [talk]
Neil Shah (UC Berkeley, Barracuda Networks)*; Grant Ho (UC Berkeley, Barracuda Networks); Marco Schweighauser (Barracuda Networks); Mohamed Ibrahim (Barracuda Networks); Asaf  Cidon (Columbia University); David Wagner (UC Berkeley)
MALOnt: An Ontology for Malware Threat Intelligence [talk]
Nidhi Rastogi (Rensselaer Polytechnic Institute)*; Sharmishtha  Dutta (Rensselaer polytechnic institute); Mohammed Zaki (RPI); Alex Gittens (RPI); Charu Aggarwal (IBM)
09:35am – 09:45amBreak
09:45am – 10:15amTalks + Q&A II: Adversarial ML for Better Security (Session Chair: Arridhana Ciptadi)
FraudFox: Adaptable Fraud Detection in the Real World [talk]
Matthew Butler (Amazon)*; Yi Fan (Amazon); Christos Faloutsos (CMU)
Towards Practical Robustness Improvement for Object Detection in Safety-critical Scenarios [talk]
Zhisheng Hu (Baidu USA)*; Zhenyu Zhong (Baidu USA)
Domain Generation Algorithm Detection utilizing Model Hardening through GAN-generated Adversarial Examples [talk]
Nathaniel Gould (Georgia Institute of Technology)*; Taishi Nishiyama (NTT); Kazunori Kamiya (NTT)
10:15am – 10:45amTalks + Q&A III: Threats on Networks (Session Chair: Gang Wang)
Toward Explainable and Adaptable Detection and Classification of Distributed Denial-of-Service Attacks [talk]
Yebo Feng (University of Oregon)*; Jun Li (University of Oregon)
Forecasting Network Intrusions from Security Logs Using LSTMs [talk]
Graham Mueller (Leidos)*; Alex Memory (Leidos); Kyle Bartrem (Leidos)
DAPT 2020 – Constructing a Benchmark Dataset for Advanced Persistent Threats [talk]
Sowmya Myneni (Arizona State University)*; Ankur Chowdhary (Arizona State University)*; Abdulhakim Sabur (Arizona State University); Sailik Sengupta (Arizona State University); Garima Agrawal (Arizona State University); Dijiang Huang (Arizona State University); Myong Kang (US Naval Research Lab)
10:45am – 11:50amPanel Session
Sadia Afroz (ICSI and Avast Software)
Cormac Herley (Microsoft Research) 
Sean Peisert (Berkeley Lab and UC Davis)
Moderator: Ali Ahmadzadeh
11:50am – 12:00pmClosing Remark

Each talk + Q&A session will begin with short presentations for all papers in the session (3 minutes each), followed by a Q&A moderated by one of the chairs.

%d bloggers like this: