The project will result in the development of a computing platform and experimental workstations for the scientific community and industry in Poland and abroad. The PLAI4SCIENCE research agenda covers the topics of:

1. ML-assisted material simulations: study of the properties of molecules and nanostructures; study of the properties of low-dimensional optoelectronic systems; development and use of quantum-chemical and AI/ML-assisted simulation methods to reduce the cost of theoretical calculations and enable simulation of large systems that are difficult to process with currently available quantum-chemical methods..
   • Aplikacje:
     • predicting the properties of multi-electron systems,
     • computational chemistry,
     • spectroscopic calculations,
     • materials engineering
     • molecular dynamics,
     • Design of medicines,
     • material identification for photovoltaics, spintronics and organic electronics industries.
2. Molecular spectroscopy and photonic metrology: use of optical resonance cavities, ultraprecision spectroscopy, optical frequency combs to measure material properties and ultrafast processes and validate spectroscopic models calculated using AI methods and ML models, ‘smart’ light sources..
   • Aplikacje:
     • characterisation of materials for the semiconductor and optoelectronics sectors,
     • generation of reference data for atmospheric monitoring and trace detection systems,
     • monitoring of technological processes,
     • biomedical diagnostics,
     • precise characterisation of laser systems.
3. Measurements using spatial-spectral imaging: hyperspectral imaging with ML models for detection, segmentation and classification of spectra and dedicated computer vision models.
   • Aplikacje:
     • environmental and industrial monitoring,
     • quality control (e.g. food quality),
     • non-contact detection and identification of substances,
     • medical diagnostics.
4. Applications of explainable AI and ML methods in the sciences: specialised algorithms and models, both classical and deep neural network architectures, e.g. graph networks and language models, as well as tools for model training and reinforcement learning. An advanced computing environment with high-powered clusters and appropriate software is part of the infrastructure.