BEGIN:VCALENDAR
VERSION:2.0
PRODID:https://evenements.uqam.ca
X-PUBLISHED-TTL:P1W
BEGIN:VEVENT
UID:34637@https://evenements.uqam.ca
DTSTART:20261203T103000Z
SEQUENCE:4
TRANSP:OPAQUE
URL:https://evenements.uqam.ca/evenements/seminaire-au-dic-multimodal-evalu
 ation-state-computation-par-yoav-artzi/34637?date=2026-12-03_10-30-00
LOCATION:UQAM - Pavillon Président-Kennedy (PK) (201\, avenue du Présiden
 t-Kennedy\, Montréal )
SUMMARY:Séminaire au DIC: «Multimodal evaluation/State computation» par 
 Yoav Artzi
CLASS:PUBLIC
DESCRIPTION:Séminaire ayant lieu dans le cadre du Doctorat en informatique
  cognitive\, en collaboration avec le centre de recherche CRIA      
         \n\n\n \n\n\nTITRE : Multimodal evaluation/State computa
 tion\n\n\n \n\n\nYoav ARTZI\n\n\nJeudi le 3 décembre 2026 à 10h30\n\n\n
 Local PK-5115 (Il est possible d'y assister en virtuel en vous inscrivant
  ici)                 \n\n\n \n\n\nRÉSUMÉ\n\n\nThis ta
 lk briefly describes three distinct projects. The first two focus on evalu
 ating multi-modal capabilities in models through the evaluation of dual pe
 rspective reasoning and fine-grained manipulation. For the former\, we foc
 us on spatial reasoning across egocentric and allocentric observations\, f
 ollowing setups designed for the evaluation of cognitive development in ch
 ildren. The latter utilizes knot manipulation\, a scenario well modeled by
  mathematical theory\, which provides a clear avenue to gauge complexity a
 nd generalization. The last part of the talk is focused on a fundamental c
 apability critical for cross-modal reasoning: the ability to separate betw
 een how one represents state and one's prediction of outputs. We hypothesi
 ze that mechanically separating the two in LLMs is beneficial and show thi
 s in the context of LLM pre-training.\n\n\n \n\n\nBIOGRAPHIE\n\n\nYoan AR
 TZI is Associate Professor in the Department of Computer Science and Corne
 ll Tech at Cornell University and visiting research faculty at Google Deep
 Mind. His research is on language modeling and learning in interactive and
  situated scenarios.\n\n\n \n\n\nRÉFÉRENCES \n\n\nChen\, Z.\, &amp\; A
 rtzi\, Y. (2025). Knot so simple: A minimalistic environment for spatial 
 reasoning. Advances in Neural Information Processing Systems: Datasets an
 d Benchmarks Track (NeurIPS DB). arXiv \n\n\nMonea\, G.\, Godey\, N.\, B
 rantley\, K.\, &amp\; Artzi\, Y. (2026). The state-prediction separation 
 hypothesis. arXiv. arXiv  \n\nMot-clés : Séminaire\, LLMs\, NeuroQAM\
 , Neurolab\, LATECE UQAM INFORMATIQUE\, LATECE\, CRIA\, Département de Ne
 uroscience\, Sciences cognitives\, Philosophie\, Sciences cognitive\, Éco
 le de langues\, neurosciences cognitives\, Institut des sciences cognitive
 s\, Apprentissage du langage naturel\, sciences du langage\, apprentissage
  machine\, apprentissage profond\, langage automatique\, langage cognitif\
 , metacognition\, Cognition humaine\, Cognition\, intelligence artificiell
 e\, IA\; intelligence artificielle\; société\, IA \; intelligence artifi
 cielle\, société\, IA\, intelligence artificielle\, chatGPT\, enseigneme
 nt supérieur\, IA\, intelligence artificielle\, département de linguisti
 que\, doctorat en psychologie\, maîtrise en psychologie\, Maîtrise en in
 formatique\, département de psychologie\, Faculté des sciences humaines\
 , Faculté des sciences de l'UQAM\, Faculté des sciences\, Département d
 'informatique\, doctorat en informatique cognitive\, doctorat en informati
 que\n\nPrix : Gratuit\n\n
CATEGORIES:Séminaire,Conférence
DTSTAMP:20261023T082618Z
CREATED:20261014T184126Z
LAST-MODIFIED:20261015T110643Z
END:VEVENT
END:VCALENDAR