Annual Computer Security Applications ConferenceJun 20, 2024
For this #ThrowbackThursday, we will look at #ACSAC2023's Machine Learning Security I session. The links in this thread will lead you to the paper pdfs and the slide decks, so be sure to check them out! 1/4
Show threadAnnual Computer Security Applications ConferenceJun 20, 2024
Opening the session was Zheng et al.'s "Secure Softmax/Sigmoid for Machine-learning Computation," exploring numerical approximations to improve efficiency and accuracy in secure computation. (https://www.acsac.org/2023/program/final/s29.html) 2/4
#MachineLearning #SecureComputation #MultipartyComputation
ACSAC2023 Program – powered by OpenConf

Show threadAnnual Computer Security Applications Conference
Second in the session was Wang & Wang's "Link Membership Inference Attacks against Unsupervised Graph Representation Learning," which investigates privacy vulnerabilities in UGRL models. (https://www.acsac.org/2023/program/final/s57.html) 3/4
#UnsupervisedLearning #PrivacyInAI #GraphAnalytics
ACSAC2023 Program – powered by OpenConf

Jun 20, 2024 at 2:00pmWeb
Show threadAnnual Computer Security Applications ConferenceJun 20, 2024
Last but not least, came Tekgul & Asokan's "FLARE: Fingerprinting Deep Reinforcement Learning Agents using Universal Adversarial Masks" which is robust to model modification attacks. (https://www.acsac.org/2023/program/final/s264.html) 4/4
#MachineLearningSecurity #DeepReinforcementLearning #SecurityInAI
ACSAC2023 Program – powered by OpenConf