🤖🛠️ Multimodal LLMs are getting better at reasoning, but what about their robustness against adversarial images? Our latest paper (which will be presented tomorrow at COLM 2024!) tackles this question through Chain-of-Thought (CoT) reasoning. A thread 🧵:
It was the most important machine learning conference of that decade and is leading. We were shallower, but also deeper. Thanks to my co-authors: R. Hofmann, M. Roescheisen, J. Hollatz, S. Ahmad, R. Neuneier, M. Taniguchi, D. Ormoneit, H. G. Zimmermann, T. Briegel, J. Hollmen
On a historical note: During the Golden Decade of the Conference on Neural Information Processing Systems (then NIPS, now, NeurIPS) from 1990-2000, I had 16 papers published in its proceedings, more than any researcher working in Europe.
We got a lot of positive feedback during the workshop. Thanks to everyone attending and also to ELLIS for providing the opportunity to realize this event in Oberwolfach.
Three days of intense discussions & expert talks: Our #ELLISProgram ‘Semantic, Symbolic and Interpretable #MachineLearning’ gathered a group of international researchers for a workshop at the Oberwolfach Research Institute for Mathematics in Germany!
👉 https://t.co/YfhaUqxdox
Registration for general attendance now open for this year's mathematics of neuroscience conference:
https://t.co/pJDtURWsZM
We've got a full schedule of great talks, and would love to have you join us!
Also accepting physical/virtual poster submissions until Sept 20th.
Registration for general attendance now open for this year's mathematics of neuroscience conference:
https://t.co/pJDtURWsZM
We've got a full schedule of great talks, and would love to have you join us!
Also accepting physical/virtual poster submissions until Sept 20th.
If one agent sees an image and utters: "A yellow and black bird is on tree branch" how do we know that another agent understands this correctly? This is studied in our ICCV paper (1/3) "Do DALL-E and Flamingo Understand Each Other?" https://t.co/nQNrZaRBFF
This workshop proved to be a true collaboration of various disciplines, bringing together researchers from neuroscience, psychology, philosophy, cognition, and AI. I express my gratitude to Sen Cheng from Ruhr-Universität Bochum and all the organisers.
Without memories we have no understanding of the world, or as Goethe put it: “We only see what we know.“ My student Hang Li presented our work on the tensor brain https://t.co/1zRscCDEjP at this https://t.co/qyLdfnmb58 workshop on generative episodic memories.
Nicht nur auf Risiken von generativer #KI schauen, sondern Chancen in Deutschland nutzen, mit Rechenleistung und Innovationsgeist, so Nicolas Flores-Herr @FraunhoferIAIS und @vtresp auf der #PLSKonferenz2023
Honored and happy to (re-)join my alma mater
@LMU_Muenchen as honorary professor for #MachineLearning in the #LifeSciences, 16 years after completing my doctoral studies in Munich