Tobias Leemann
@t_leemann
PhD Student in Explainable, Private and Reliable ML | @uni_tue @TU_Muenchen
You might like
Differential privacy is a hammer, but not every privacy problem is a nail. 🔨 We introduce "Gaussian Membership Inference Privacy" in our #NeurIPS2023 paper with @MartinPawelczyk and @Gjergji_. A thread 🧵👇 [1/n]
![t_leemann's tweet image. Differential privacy is a hammer, but not every privacy problem is a nail. 🔨
We introduce "Gaussian Membership Inference Privacy" in our #NeurIPS2023 paper with @MartinPawelczyk and @Gjergji_.
A thread 🧵👇 [1/n]](https://pbs.twimg.com/media/F-5ppakWQAAN0wR.png)
Auto-GDA: Automatic Domain Adaptation for Efficient Grounding Verification in Retrieval Augmented Generation Presents a framework for unsupervised domain adaptation of lightweight NLI models used in RAG systems. 📝arxiv.org/abs/2410.03461
Happy to be on the list! ☺️
[LG] Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers arxiv.org/abs/2405.13536 - Common transformer architectures like BERT and GPT-2 structurally cannot represent additive models like linear models or generalized additive…
![fly51fly's tweet image. [LG] Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers
arxiv.org/abs/2405.13536
- Common transformer architectures like BERT and GPT-2 structurally cannot represent additive models like linear models or generalized additive…](https://pbs.twimg.com/media/GOYG0ZTbcAAS1ze.jpg)
![fly51fly's tweet image. [LG] Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers
arxiv.org/abs/2405.13536
- Common transformer architectures like BERT and GPT-2 structurally cannot represent additive models like linear models or generalized additive…](https://pbs.twimg.com/media/GOYG0ZRbgAAc2Z4.jpg)
![fly51fly's tweet image. [LG] Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers
arxiv.org/abs/2405.13536
- Common transformer architectures like BERT and GPT-2 structurally cannot represent additive models like linear models or generalized additive…](https://pbs.twimg.com/media/GOYG0veaEAAXEqS.jpg)
🚨 Exciting #PhD opportunities in #DataScience and #ResponsibleAI @TU_Muenchen @MdsiTum 🎓 Passionate about research in responsible AI and Data Science? Join our talented team for groundbreaking research and innovation #GenerativeAI #LLMs #AIRegulation portal.mytum.de/jobs/wissensch…
Is Claude not available in the EU? Why??
I am very happy to celebrate my PhD defense. I want to take this opportunity to share my unique path through my PhD and share a small conclusion.
📢 Please RT!📢 We're hiring postdoctoral researchers to work on responsible machine learning at UCSD! Topics include fairness, explainability, robustness, and safety. For more, see berkustun.com/postdoc/
United States Trends
- 1. Araujo 176K posts
- 2. Chelsea 609K posts
- 3. Barca 245K posts
- 4. Estevao 243K posts
- 5. Barcelona 439K posts
- 6. Ferran 75.6K posts
- 7. Yamal 187K posts
- 8. Oklahoma State 4,419 posts
- 9. Eric Morris 2,923 posts
- 10. Cucurella 94.8K posts
- 11. Flick 46.2K posts
- 12. Godzilla 24.8K posts
- 13. Rashford 24K posts
- 14. Kounde 47.9K posts
- 15. Pedri 51.7K posts
- 16. Enzo 65.9K posts
- 17. Skippy 4,335 posts
- 18. #CHEBAR 16.5K posts
- 19. Raising Arizona 1,704 posts
- 20. Happy Thanksgiving 24.8K posts
Something went wrong.
Something went wrong.