Melis İlayda Bal
@melisilaydabal.bsky.social
PhD student @Max Planck Institute for Intelligent Systems | working on optimization & game-theoretic frameworks for robust and efficient ML | organizer @twiml.bsky.social | https://melisilaydabal.github.io
🌺Excited to share our #ICLR2025 paper: 𝗔𝗱𝘃𝗲𝗿𝘀𝗮𝗿𝗶𝗮𝗹 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗳𝗼𝗿 𝗗𝗲𝗳𝗲𝗻𝘀𝗲 𝗔𝗴𝗮𝗶𝗻𝘀𝘁 𝗟𝗮𝗯𝗲𝗹 𝗣𝗼𝗶𝘀𝗼𝗻𝗶𝗻𝗴 𝗔𝘁𝘁𝗮𝗰𝗸𝘀!
Training ML models on public data isn’t all sunshine and rainbows—especially when adversaries sneak in poisoned labels to mess with your model. But fear not! In our work, we introduce 𝗙𝗟𝗢𝗥𝗔𝗟🌺...
Training ML models on public data isn’t all sunshine and rainbows—especially when adversaries sneak in poisoned labels to mess with your model. But fear not! In our work, we introduce 𝗙𝗟𝗢𝗥𝗔𝗟🌺...
February 28, 2025 at 9:56 PM
🌺Excited to share our #ICLR2025 paper: 𝗔𝗱𝘃𝗲𝗿𝘀𝗮𝗿𝗶𝗮𝗹 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗳𝗼𝗿 𝗗𝗲𝗳𝗲𝗻𝘀𝗲 𝗔𝗴𝗮𝗶𝗻𝘀𝘁 𝗟𝗮𝗯𝗲𝗹 𝗣𝗼𝗶𝘀𝗼𝗻𝗶𝗻𝗴 𝗔𝘁𝘁𝗮𝗰𝗸𝘀!
Training ML models on public data isn’t all sunshine and rainbows—especially when adversaries sneak in poisoned labels to mess with your model. But fear not! In our work, we introduce 𝗙𝗟𝗢𝗥𝗔𝗟🌺...
Training ML models on public data isn’t all sunshine and rainbows—especially when adversaries sneak in poisoned labels to mess with your model. But fear not! In our work, we introduce 𝗙𝗟𝗢𝗥𝗔𝗟🌺...