Idiap Symposium part #2 - Perspectives on AI

As part of its symposium, theIdiap Research Institute, the initiator of the event, announces the 2nd part of its series of 4 RDV dedicated to artificial intelligence: " Perspectives on AI" . These events aim to bring together academics, industrialists and policy-makers to present their views on AI, in particular on how we can leverage it to tackle major societal and industrial challenges. This 2nd instalment will focus on deep learning.

Online registration: ? 5 free places are available for the Alp ICT community - contact the team to take advantage!

Beyond Deep Learning: Learning from Small and Heterogeneous Data

Over the past decade, deep learning has become the predominant paradigm in artificial intelligence thinking. However, there are many relevant scenarios in which traditional deep learning methods are not suitable, when large volumes of data are not available to train models, or when this data is too heterogeneous and multimodal (i.e. involves a combination of several means of communication between the user and the machine/interface), for example. Added to this is the need for many critical application domains that make data interpretation a central issue. Examples where these requirements are the norm are numerous: from clinical trials to policy-making based on the analysis of complex data on human interactions, behaviors or the environment.

In this second installment of Perspectives on AI, emerging methods that are not part of the traditional deep learning toolbox and that aim to meet these challenges will be discussed. Under the theme "Doing better with less", speakers from academia and industry will discuss how recent methodological developments in AI and statistics can provide solutions to existing challenges and offer new opportunities for making sense of an increasingly vast and complex data horizon. 

Program 

• 9:30 - 9:40 : Welcome - Dr Lonneke Van Der Plas, Group Leader at Idiap, Computation, Cognition & Language
• 9:40 - 10:30 : Integrating diverse evidence: from science to policy-making - Prof. Elizabeth Tipton, Northwestern University
• 10:30 - 10:55 : There is more to learning than data - Prof.Alexandros Kalousis, University of Applied Sciences
• 10:55 - 11:20 : Uncertainty reduction strategies over indirect, sparse data - Cedric Travelletti, Executive Director at Julius Baer
• 11:45 - 12:10 : AbDiffuser: full-atom generation of in-vitro functioning antibodie - Dr Andreas Loukas, Prescient Design, Genentech/Roche
• 12:10 - 12:35 : to be defined - Dr Dorina Thanou, Senior research scientist, EPFL
• 12:35 - 13:15 : Panel: The future of data science (Challenges & Opportunities) - Prof. Elizabeth Tipton, Dr Alexandros Kalousis, Dr Cedric Trevelleti & Dr Andreas Loukas, Dr Dorina Thainou
• 14:15 - 14:45 : Rich data: heterogeneity, lack of annotations, high dimensionality - Prof. Matteo Valleriani, Max Planck Institute
• 14:45 - 15:15 : Matching ML solutions with organizations: a perspective from the Swiss Data Sc - Silvia Quarteroni, Swiss Data Science Center
• 15:15 - 15:45 : The might of abduction: integrating textual and numerical evidence - Dr André Freitas, Idiap Research Institute
• 15:45 - 16:25 : Industrial Panel: Opportunities for economic & social impact
• 16:25 - 16:30 : Closing remarks - Dr Raphaëlle Luisier, Group Leader at Idiap, SIB Swiss Institute of Bioinformatics

Online registration: ? 5 free places are available for the Alp ICT community - contact the team to take advantage!