ANITI, Artificial and Natural Intelligence Toulouse Institute, is the interdisciplinary artificial intelligence institute of Toulouse. Held by the University of Toulouse, is based on 3 main pillars: scientific research, training and contribution to economic development.
It is housed in TGE B612 builiding and hosts around 200 researchers from universities, engineering schools and scientific and technological research organizations in Toulouse and its region, as well as around 50 partners.
The Institute has targeted as strategic application sectors: (1) intelligent mobility and transport, and (2) robotics/cobotics for the industry of the future. ANITI aims to promote exchanges between industrialists who address their problems to scientists and academics in order to develop new markets and new technological applications.
ANITI is the coordinator of TUPLES and contributes its expertise in automated planning & scheduling, machine learning, explanations, and trustworthy AI to the project in the following ways:
- First – Project Manager Naomi Wolff and Project Coordinator Sylvie Thiebaux lead Workpackage 2, which is the coordination workpackage.
- Second – In Workpackage 3, ANITI will leverage the expertise of Dr Emmanuel Hébrard and Prof Sylvie Thiébaux to develop new hybrid methods for planning and scheduling that combine search or constraint programming with learnt heuristics and policies to scale up to larger problems.
- Third – In Workpackage 5, which focuses on explanations, Prof. João Marques-Silva‘s expertise on formal explanations for machine learning models will be used to extend these techniques to be more interactive and efficient, and to handle the more complex case of models representing plans, schedules and policies in collaboration with Saarland University.
- ANITI will also investigate the ethical and legal aspects of explanations, leveraging the expertise of Profs. Nicholas Asher and Céline Castets-Renard.
- Finally, in Workpackage 6, ANITI will continue the ongoing collaboration with Airbus on hybrid methods for task scheduling in manufacturing, and collaborate with SciSports on explanation techniques for the sports use case as well as with Optit on the challenging future energy systems use case.
Three ANITI Chairs are involved in TUPLES: first, the Chair on Deep Learning Verification (DeepLEVER), headed by João Marques-Silva, which aims at explaining and verifying machine learning systems via combinatorial optimization; second, the Chair on Law, Accountability, and Social Trust in AI, headed by Céline Castets-Renard, which studies the social risks that AI might generate; and third, the Chair on Automated Planning, headed by Sylvie Thiébaux, which aims at building safe, robust, and efficient hybrid planning systems that combine the advantage of reasoning and learning.
For further information: https://youtu.be/6T5PcJxMjiE
- strategic areas of application : mobility/transport & robotic/cobotic
- +200 researchers
- 3 integrative programs
- 24 scientific chairs
- 1 graduate school
- +50 partners
#tuples #tuplesai #aniti #trusthwortyai #responsibleai #explainableai #ai #horizoneu
We are very pleased to announce that our coordinator, Sylvie Thiebaux, is at the prestigious IJCAI23 conference both as Area Chair and as presenter of the paper "Formal Explanations of Neural Network Policies for Planning".
The workshop for Hierarchical Planning (HPlan) was founded in 2018 and is carried out since then every year at the International Conference of Automated Planning and Scheduling (ICAPS). Among the Organizing Committee of this 2023 edition is Daniel Höller from Saarland University, member of the TUPLES project.
The 2023 ICAPS (International Conference on Automated Planning and Scheduling) featured an array of captivating workshops, and one that involved many TUPLES’ members was the Workshop on Robust Decision Making for Planning and Scheduling (RDDPS).
Rebecca Eifler and Jörg Hoffmann from TUPLES are members of the organizing and steering committees respectively. Daniel Holler, member of the TUPLES Consortium, will give an invited talk.