Menu Fermer

“Save The Date” – Séminaire IA hybride : applications et défis – Mercredi 13 septembre – LIRMM


Ce séminaire est organisé dans le cadre des axes transverses “IA et science des données” du LIRMM.

Le Séminaire Hybrid AI: Applications and Challenges qui aura lieu le mercredi 13 septembre 2023 de 9h30 à 14h au LIRMM.

Voici un pré-programme et plus d’info sur les invités :

– 9h30 Welcome & opening – Konstantin Todorov, MCF LIRMM / Unviersity of Montpellier
– 10h00 Achim Rettinger, Full Professor for AI at the University of Trier (Germany): talk + discussion
– 10h45 Pause café
– 11h00 José Manuel Gomèz Perez, CEO and co-founder of Expert.AI (Spain): talk + discussion
– 11h45 Round table on hybrid AI, interdisciplinary approaches, applications and challenges
– 12h30 – 14h00 Buffet

Achim Rettinger, Full Professor for AI at the University of Trier, leader of the Knowledge Representation Learning group ( and head of FZI, the Computer Science Research Center (

Title: Hybrid AI Systems for Transfer Learning using Knowledge Graphs

Transfer Learning (TL) is an area that has been core to machine learning research since before the deep learning hype. It is about generalizing a machine learning system to domains or tasks it has not been specifically trained for. With the rise of deep learning and embeddings considerable progress could be achieved in this area. In this talk I will focus on how knowledge graphs can be utilized to improve deep learning models. The primary application domain of hybrid AI systems proposed in this talk is visual object recognition. The talk will be concluded with the open research question on how foundation models and in-context learning contribute to transfer learning in general and what the role of knowledge-based hybrid deep learning systems might be in particular.

BIO: Prof. Rettinger’s interests lie in developing and applying deep learning language models for the extraction of information for building large knowledge graphs in different fields, where the latter in turn can be used as a basis for inference (both statistical and symbolic) of new knowledge (i.e. predicting new relations between graph entities, clustering nodes or categorizing entire networks). Currently, Prof. Rettinger collaborates strongly with LIRMM and the UM in building a Horizon EU project on AI for fighting disinformation.

José Manuel Gomèz Perez, CEO and co-founder of Expert.AI, a Madrid-based SME for AI building and using knowledge graphs and machine learning models (

Title: The Tortuous Road to Detecting and Countering Deception

Deception, or the intent to deceive is at the root of many of the grand challenges we are faced with as a society nowadays. From misinformation and disinformation to online manipulation and radicalization, deception always plays a central role. The question is, how can we identify deception and counter it? In this talk I will go through several approaches ranging from evidence-based fact checking to understanding the nuances of deceptive language and their application to different scenarios. Given the complexity of the challenge, we will touch upon different techniques involving symbolic, neural, and hybrid AI, language understanding, and knowledge graphs.

BIO: Dr. Peréz, a former Marie-Curie fellow, has a strong academic record in the field of AI and Semantic Web. Before, he worked as Director R&D at iSOCO, one of the first European companies to deliver semantic and natural language processing solutions on the Web. His current interests lie in bringing common sense knowledge (coming from structured knowledge resources) to statistical language models like BERT and GPT-3, explaining AI models and understanding and monitoring bias in data and algorithms.

L’inscription est gratuite, mais nous vous demandons de bien vouloir le faire afin notamment de prévoir les pauses et déjeuner :…-ckxmpndi

Nous comptons sur votre participation !

Plus d’infos : Konstantin Todorov (

En voir plus