The research at PRAIRIE Institute

Research in PRAIRIE is organized in a double helix with two intertwined threads: (a) core AI methodological research, and (b) interdisciplinary work at the interface with sciences and applications.

Logo_Recherche

The first thread includes foundational work in core AI domains such as knowledge representation, machine learning and optimization, as well as integration fields such computer vision, natural language and speech understanding, and robotics. The second thread expands the traditional view of an AI centered on computer science and applied mathematics to include methodologies inspired from–and applied to– biology, cognitive science and physics, as well as social sciences and humanities. It also leverages the expertise of the PRAIRIE partners to investigate fundamental research in AI, applied to problems in healthcare, transportation, the environment, and the wealth of other domains impacted by artificial intelligence.

Fundamental research

Core methodological research :

  • Key issues include scale, reliability, and explainability
  • Key scientific challenges include distributed nonconvex optimization, few-shot learning, weakly- and un-supervised visual recognition, domain adaption for personalized text/speech understanding, reliability of robotics systems

Interdisciplinary research

It will tackle crucial problems in modeling complex systems in cognitive science, life sciences, medicine and physics, using a unique combination of large-scale datasets, novel machine learning approaches and expert scientific knowledge to address problems such as selecting the pertinent information in high-throughput genomic and imaging data, developing computer-aided clinical decision/diagnosis systems, and modeling human cognition. This will result in both radical advances in our understanding of living organisms and high-impact applications in drug discovery and personalized medicine for example. 

Interdisciplinary efforts will play a key role in research at Prairie in areas such as:

  • AI and biology: from deep learning for single-molecule microscopy to biologically inspired artificial neural architectures.
  • AI and cognitive science: integrating AI techniques with social and cognitive sciences studies to obtain powerful predictive models of individual and collective human behavior.
  • AI and digital humanities: from the analysis of astronomy historical tables to the preservation of endangered cultured heritage.
  • AI and medicine: building advanced computer-aided decision and diagnosis systems for personalized medicine, including the evaluation, fairness, transparency, and explicability issues encountered there.
  • AI and physics: from theoretical models of learning processes inspired by statistical physics to applications of machine learning to cosmology and Earth science problems.
  • AI and social sciences: studying the impact of AI on industry, finance, and the governance of public bodies, including the corresponding political, social, legal and ethical issues.

Collaborative research with industry

  • New problems, sources of data, opportunities for transfer & innovation
  • An opportunity to confront omnipresent legal, ethical, and regulatory issues
  • Blue-sky research and more focused projects
  • High-impact applications, including environment, health and transportation