We should also expell dual citizens with Israel. As well as any political official with a dual citizenship. Just to be safeml.

A very inspiring story from Christian Szegedy - believe in your research ;) Adversarial examples are important, and 🔥 defending 🔥 against adversarial attacks is equally so. Check out our poster at #111 (Thursday 16:30-18:30), Randomized Feature Defense Against Adversarial Attacks (github.com/mail-research/ran…), a 💪defense 💪 that is: - pluggable (can be used with any already trained model or other defenses) - lightweight (almost same inference time as of original model) - training-free (not require any training modification) #safeml #adversarialtraining #ICML2024

Interested in **same-inference-time adversarial defenses** for continuous (**image**) or discrete (**text**) data? Check out our recently accepted 🙂 (openreview.net/forum?id=E296…, congrats @hungquang9999) and rejected (openreview.net/forum?id=vZ6r…, next time @DuyCao23350) 😅 #ICLR2024 papers. On the rejection: good reviews (from responsive reviewers). AC: novel, neat, and clever; also AC: rejection for reasons not discussed or already addressed 😬😬😬 in the rebuttal. Perhaps, there should be “meta-rebuttal-phase”? Nah, it’s lots of work LOL. As my collaborator (great guy) often said: “A great piece of work doesn’t vanish just because of one review/venue; instead, it lasts forever.” So, keep up the good work, Duy! #adversarialML #SafeML

For us who work on #Backdoor (and also #safeML research), do you think these attacks will manifest in life? Interesting question from @FlorianTramer! #BUGS @NeurIPSConf #NeurIPS2023

I am glad to announce that we have extended the idea SafeML for Timeseries in collaboration with @FraunhoferIESE. The paper will be presented at @WAISE_AIsafety 2022. Please feel free to check our paper at arxiv: doi.org/10.48550/arXiv.2206.… GitHub Repository: github.com/n-akram/TimeSerie…

5/ 3] In “Maximum Weighted Loss Discrepancy” [SafeML, ICLR 2019] paper, they measure the loss discrepancy for an exponential number of groups arxiv.org/pdf/1906.03518.pdf

Going live in next 4hrs. Join us at our @YouTube live event @PyData @Pune event youtu.be/4z_CgpJ_x_c @adversarial_nik @NumFOCUS #safeguardingml #safeml

I would like to share our codes for SafeML project in Github. The project is in its early stages and it should be improved a lot. Please feel free to check our SafeML Github repository: github.com/ISorokos/SafeML #SafeML #AIsafety #DeepLearning #SafeAI #SafeDl #Safety

SafeML: Safety Monitoring of Machine Learning Classifiers through Statistical Difference Measure. (arXiv:2005.13166v1 [cs.LG]) arxiv.org/abs/2005.13166

SafeML: Safety Monitoring of Machine Learning Classifiers through Statistical Difference Measure deepai.org/publication/safem… by @koo_ec et al. #CumulativeDistributionFunction #MachineLearning

SafeML: Safety Monitoring of Machine Learning Classifiers through Statistical Difference Measure. (arXiv:2005.13166v1 [cs.LG]) ift.tt/3dc1vpd

We're organizing the Safe Machine Learning Workshop #safeml at @ECAI2020 Submission deadline: 15th March safeml.bitbucket.io/ We're looking forward to your papers! 😉

‼️Pues parece que es oficial!! @moisipm y yo co-organizaremos el workshop SafeML para @ECAI2020. Pronto más info 👇👇 This workshop aims to bring together papers outlining the safety and fairness implications of ML in real-world systems. Topics include but are not límited to:

Very excited to deliver the #icml2019 tutorial on #safeml tomorrow together with @csilviavr! Be prepared for fairness, human-in-the-loop RL, and a general overview of the field. And lots of memes!

Separate recordings of SafeML talks are now available on the workshop website sites.google.com/corp/view/s…

Thanks to those who came to the #safeML poster session and checked out our paper on Using Videos to Evaluate Image Model Robustness. You can find our paper @ arxiv.org/abs/1904.10076 with Brandon Yang @quocleix @JiquanNgiam and Jon Shlens. #ICLR #ICLR2019

Recordings of SafeML workshop talks are available on the #ICLR2019 live stream website (fast forward to 26:40) slideslive.com/38915487/r06-…

Researchers come to the topic of ethical AI from different vantage points - social justice and high performance ⁦@dhadfieldmenell⁩ #SafeML #ICLR2019

Presenting joint work between @CognitiveScale and @UTAustin that delivers realistic counterfactual explanations to users and also allows model developers to study fairness and robustness at #SafeML #ICLR2019 please stop by!

Come to our posters at #ICLR2019 if you want to chat: 1. Mon 10:30 Generalizing from a few environments in safety-critical reinforcement learning (SafeML Workshop, Room R6) 2. Wed 4.30pm An Empirical study of Binary Neural Networks' Optimisation (Great Hall BC #14) @OATML_Oxford