Perfect #trainingdata doesn't exist to learn all 24 types of #syllogism. Instead, it is possible to learn all syllogism types by learning only semantic representations of the input. Find out more in our latest #Blog post by #ML2R's Tiansi Dong (Rheinische Friedrich-Wilhelms-Universität Bonn): machinelearning-blog.de/forschung/syll…

Die Hochschulwahlversammlung der Universität Bonn hat soeben vier Prorektorinnen und zwei Prorektoren gewählt. Das Rektorat ist künftig geschlechterparitätisch besetzt.

How can #computers represent the meaning of words when its spelling is not a sufficient indicator? Through the use of #numerical #vectors that include the neighboring words, explains Gerhard Paass (Fraunhofer IAIS) in the 🇩🇪 #ML2R #Blog: ➡️ machinelearning-blog.de/grundlagen/kon…

#ML2R's Tiansi Dong is lead organizer of a prestigious Schloss Dagstuhl Seminar (5-10 Sept.). With interdisciplinary experts from Universität Trier Tsinghua University Columbia University & Vrije Universiteit Amsterdam (inactief), he addresses the integration of structure & learning in #MachineLearning & #AI. ➡️ dagstuhl.de/21362

The Schloss Dagstuhl Seminar 'Structure and Learning' - lead organizer: #ML2R's Tiansi Dong - concluded on Friday with presentations by the seminar's working groups. Several collaborations on a neural-symbolic approach to #AI have been established as a result of the weeklong event.

Semifinals testing at the ANA Avatar Xprize Challenge in Miami. The operator is telepresent in the NimbRo Avatar robot. nimbro.net/AVATAR

The second and last testing day at the ANA Avatar Xprize Challenge Semifinals also went very well. We may have received even more points than yesterday. nimbro.net/AVATAR

#BoxingDay is: now!📦 🗣️ Our #ML2R #AutumnSchool kicks off today at 9.30 a.m. CEST. During this week, over 30 selected #MachineLearning enthusiasts (10 countries) will work & learn together to apply their #ML solutions to a concrete #BinPacking problem. ℹ️ ml2r.de/en/autumn-scho…

We are happy to welcome another selected keynote speaker, Tiansi Dong Rheinische Friedrich-Wilhelms-Universität Bonn @MLCompetence, at our #AutumnSchool in two weeks! Click here to register for free: 👉events.scads.ai/event/4/ TU Dresden ZIH (TU Dresden) UNIVERSITÄT LEIPZIG @RALeipzig #NHR @InfAI_eV futureSAX DRESDEN-concept

Want to know why it's risky to assume that your ML is going to continue work if the test regime changes from your training data? Just ask Zillow. wired.com/story/zillow-i…

We had 3 great days at our #AutumnSchool with amazing people and fascinating talks about #AI, #BigData, #MachineLearning... Thank you for your participation🙏🏽See you next year!👀 TU Dresden ZIH (TU Dresden) SchauflerLab@TU Dresden UNIVERSITÄT LEIPZIG DRESDEN-concept Tiansi Dong zeuchnes Helge Holzmann

The 🆕 report of the Schloss Dagstuhl seminar "Structure and Learning" in #AI, organized by #ML2R researcher Tiansi Dong (Rheinische Friedrich-Wilhelms-Universität Bonn), is now available online! Learn more about the integration of symbolic & numeric #inference from an interdisciplinary perspective: doi.org/10.4230/DagRep…

Current LLMs are trained on text data that would take 20,000 years for a human to read. And still, they haven't learned that if A is the same as B, then B is the same as A. Humans get a lot smarter than that with comparatively little training data. Even corvids, parrots, dogs,

Sphere Neural-Networks (SphNNs) ... imbuing neural networks with capabilities for more complex and deterministic forms of reasoning, including syllogistic reasoning—a foundational element of human rationality. emergentmind.com/papers/2403.15…

Sphere Neural Networks (SphNNs) signify innovative qualitative enhancement to endow LLMs with the capability for high-level cognition. goatstack.ai/topics/introdu… #RationalReasoning #CognitiveModeling #Human-likeReasoning #Neuro-SymbolicAI #goatai #goatstack via GoatStack.AI

With Spatial Intelligence, AI Will Understand the Real World | Fei-Fei L... youtu.be/y8NtMZ7VGmU?si… via YouTube

Reviewer 2: ACL is not an AI conference. Reject.

#NewProfilePic

Neural Reasoning and Mathematical Discovery – An Interdisciplinary Two-Way Street (NeurMAD@AAAI’25) neurmad.github.io

Four methodological limitations prevent machine learning systems from reaching the rigour of logical reasoning. They cannot, not because of insufficient training data. On the contrary, to achieve the rigour, they shall not use training data. talks.cam.ac.uk/talk/index/228…