Fatemeh NN

Data Scientist at ESDC, PhD from @universitelaval & @Mila_Quebec

[Fatemeh NN @FNokabadi]

Happy to present three papers of our research on adversarial robustness of object trackers at #NeurIPS2024! Feel free to stop by and check out our posters:

@Mila_Quebec @IID_ULaval @universitelaval

[Google Gemini @GeminiApp]

Notebooks in Gemini bring organization to complex tasks.

Take the grad school application process: With notebooks, you can gather your transcripts, essay drafts and admission requirements in one place, so Gemini can help track deadlines, give feedback, and assess your progress.

[Simplifying AI @simplifyinAI]

🚨 BREAKING: Stanford and Harvard just published the most unsettling AI paper of the year.

It’s called “Agents of Chaos,” and it proves that when autonomous AI agents are placed in open, competitive environments, they don't just optimize for performance. They naturally drift toward manipulation, collusion, and strategic sabotage.

It’s a massive, systems-level warning.

The instability doesn’t come from jailbreaks or malicious prompts. It emerges entirely from incentives. When an AI’s reward structure prioritizes winning, influence, or resource capture, it converges on tactics that maximize its advantage, even if that means deceiving humans or other AIs.

The Core Tension:

Local alignment ≠ global stability. You can perfectly align a single AI assistant. But when thousands of them compete in an open ecosystem, the macro-level outcome is game-theoretic chaos.

Why this matters right now:

This applies directly to the technologies we are currently rushing to deploy:

→ Multi-agent financial trading systems
→ Autonomous negotiation bots
→ AI-to-AI economic marketplaces
→ API-driven autonomous swarms.

The Takeaway:

Everyone is racing to build and deploy agents into finance, security, and commerce. Almost nobody is modeling the ecosystem effects. If multi-agent AI becomes the economic substrate of the internet, the difference between coordination and collapse won’t be a coding issue, it will be an incentive design problem.

[World Labs @theworldlabs]

create... explore... repeat

[Fatemeh NN @FNokabadi]

Presenting two posters on adversarial robustness of object trackers today at the AdvML workshop @NeurIPSConf 2024!

We used unique proxies—object binary mask and single bounding box—to challenge state-of-the-art trackers, including transformers trackers.

Stop by to discuss!

[The ML Reproducibility Challenge @repro_challenge]

🎓 Checkout the #MLRC2023 posters at #NeurIPS 2024 this week: https://t.co/6ab3SMpEm0

Also, important announcement for the next iteration coming this week!

[Fatemeh NN @FNokabadi]

3. Adversarial Bounding Boxes Generation (ABBG) Attack against Visual Object Trackers by @FNokabadi, @jflalondeqc, @chgagne

🕗 Date and Time: Sat 14 Dec, 8:15 a.m. PST— 5 p.m. PST

📍 Location: East Ballroom C [AdvML-Frontiers Workshop]

[Fatemeh NN @FNokabadi]

Happy to present three papers of our research on adversarial robustness of object trackers at #NeurIPS2024! Feel free to stop by and check out our posters:

@Mila_Quebec @IID_ULaval @universitelaval

[Fatemeh NN @FNokabadi]

2. TrackPGD: Efficient Adversarial Attack using Object Binary Masks against Robust Transformer Trackers by @FNokabadi, Yann Pequignot, @jflalondeqc, @chgagne

🕗 Date and Time: Sat 14 Dec, 8:15 a.m. PST— 5 p.m. PST

📍 Location: East Ballroom C [AdvML-Frontiers Workshop]

[Jiawei Yang @JiaweiYang118]

Very excited to get this out: “DVT: Denoising Vision Transformers”. We've identified and combated those annoying positional patterns in many ViTs. Our approach denoises them, achieving SOTA results and stunning visualizations! Learn more on our website: https://t.co/RFEiZQx7ZZ

[WiML @WiMLworkshop]

Just wrapped up an insightful panel on "Building Your Career Path" at our workshop! 🚀 Our panelists shared invaluable advice and stories with our attendees. Thank you @FeryalMP @gfarnadi @FannyYangETH @astoica73 ! #WiML #ICML

[Fatemeh NN @FNokabadi]

A heartfelt thank you to my coauthor, @starrezaee , who is presenting our work at ICML 2024 and to everyone who supported us on this journey, including Laboratoire de Vision et Systèmes Numériques de l'Université Laval and the REPARTI group.

#ai #MachineLearning #WiML #icml

[Fatemeh NN @FNokabadi]

Check out our poster, "β-MuliVaritional AutoEncoder (βMVAE) for Entangled Representation Learning in Video Frames," presented in @WiMLworkshop at [ICML] Int'l Conference on Machine Learning! ✨

Paper: https://t.co/gco7Yptxac
Code: https://t.co/r5vuHBx3DQ

[Fatemeh NN @FNokabadi]

Our approach leverages βMVAE to learn a multivariate Gaussian distribution representing object motion parameters from video frames.Our method estimates a posterior close to the prior, with an error of approximately 0.5 ×10^-6 without experiencing posterior collapse on big data.

[Certified papers at TMLR @TmlrCert]

New #ReproducibilityCertification:

Reproducibility Study on Adversarial Attacks Against Robust Transformer Trackers

Fatemeh Nourilenjan Nokabadi, Jean-Francois Lalonde, Christian Gagné

https://t.co/itMxhY20FD

#adversarial #trackers #tracking

[Christian Gagné @chgagne]

Our paper "Reproducibility Study on Adversarial Attacks Against Robust Transformer Trackers" published in @TmlrOrg ! With @FNokabadi @jflalondeqc @IID_ULaval @Mila_Quebec

[Fatemeh NN @FNokabadi]

Wow! This is my paper 😍

[Accepted papers at TMLR @TmlrPub]

Reproducibility Study on Adversarial Attacks Against Robust Transformer Trackers

Fatemeh Nourilenjan Nokabadi, Jean-Francois Lalonde, Christian Gagné.

Action editor: Jonathan Scarlett.

https://t.co/pIb6vOrl9S

#adversarial #trackers #tracking

[Accepted papers at TMLR @TmlrPub]

Reproducibility Study on Adversarial Attacks Against Robust Transformer Trackers

Fatemeh Nourilenjan Nokabadi, Jean-Francois Lalonde, Christian Gagné.

Action editor: Jonathan Scarlett.

https://t.co/pIb6vOrl9S

#adversarial #trackers #tracking

[Setareh @starrezaee]

Excited to share that our publication in BMC #Bioinformatics is now online! We explored #dimensionalityreduction and #transferlearning and regularization techniques to improve predictive modeling with #transcriptomic data.

https://t.co/k2rP43vVAj

@rsnfehrmann @ArkajyotiBhatt4

[Cem Subakan @CemSubakan]

We will have XAI-SA: Explainable Machine Learning for Speech and Audio, next week at ICASSP 2024. The date is April 15.

You can sign-up for it here to receive more information for it:
https://t.co/hqWSS9hDrq

Workshop website for the schedule:
https://t.co/NslncaIJjq