K. M. Graczyk

K. M. Graczyk

Laboratory of AI for Physics (LAIP)

Group of physicists keen on applications and development of AI methods for theoretical, computational, and experimental physics

  • We work on applications of AI techniques in the physics of fluids, nuclear and particle physics, and supergravity.
  • In particular, we focus on the development of:
    • Deep learning models for studies of porous media
    • Machine learning models for Monte Carlo simulations of neutrino-nucleus scattering
    • Bayesian methods for the analysis of electron and neutrino scattering data
    • Hopfield-like models
    • PINN in Bayesian approach
  • The list of papers obtained within the initiative

======

Seminar: Fridays, 10:00, room 416

2024/2025

  • 25.04.2025:
    • title: tba, speaker: tba
  • 11.04.2025:

    • title: Automated Machine Learning and Interactive Apps from Notebooks – Tools for Physicists, speaker: dr Piotr Płoński (MLJAR, Inc.)

      Abstract: In this talk, I will show how physicists can use automated machine learning (AutoML) tools to quickly build models from data. I will present MLJAR-supervised – an open-source AutoML tool that automatically creates models, selects the best one, and generates easy-to-read reports.

      I will also introduce MLJAR Studio – a user-friendly IDE that combines code, data, and AI in one place. Finally, I will demonstrate Mercury – a tool that turns Jupyter Notebooks into interactive web apps that can be shared with others.

      The talk will be practical, based on real examples, and focused on tools that make everyday research work easier.

  • 03.04.2025:
    • title: Fluid flows with LLM’s, speaker: dr hab. M. Matyka
  • 28.03.2025:
    • title: AI in Academia cont., speaker: dr R. Durka
  • 21.03.2025:

    • title: Generative adversarial neural networks for simulating neutrino interactions, speaker: dr L. Bonilla

      Abstract: We propose a new approach to simulate neutrino scattering events as an alternative to the standard Monte Carlo generator approach. Generative adversarial neural network (GAN) models are developed to simulate neutrino-carbon collisions in the few-GeV energy range. Two GAN models have been obtained: one simulating only quasielastic neutrino-nucleus scatterings and another simulating all interactions at given neutrino energy, trained on NuWro Monte Carlo simulation. The performance of both models has been assessed using two statistical metrics, and it is shown that both GAN models successfully reproduce the event distributions from the training data set.

  • 07.03.2025:

    • title: AI in Academia, speaker: dr R. Durka

      Abstract: Artificial Intelligence (AI) has the potential to reshape teaching and scientific research. We will explore real-world applications of AI tools, from improving various academic aspects and optimizing workflows to enhancing scientific exploration. I will also highlight recent developments and features, including new tools like Canva, voice-enabled AI, autonomous AI agents, deep research, and recent reasoning models.

  • 07.11.2024:
    • title: AI okiem filozofa - nieco mniej konwencjonalnie, speaker: dr Łukasz Mścisławski OP, Wroclaw University of Science and Technology
  • 24.10.2024:
    • title: Deep learning and Neutrino-Nucleus Scattering, speaker: K. Graczyk

2023/2024

  • 13.06.2024:
    • title: Attention is all you need, transformer neural network basics, speaker: B. Domański
  • 06.06.2024:
    • title: Diffusion Model for MC Simulations, speaker: L. Bonilla
  • 23.05.2024:
    • title: Diffusion by Deep Learning, speaker: D. Strzelczyk
  • 16.05.2024:
    • title: Empirical fits to inclusive electron-carbon scattering data obtained by deep-learning methods, speaker: B. Kowal
  • 09.05.2024:
    • title: Image processing in associative memory. From Hopfield networks to HopfieldLayer, speaker: B. Domański
  • 25.04.2024:
    • title: GANs for MC Generator of Events, speaker: L. Bonilla
  • 18.01.2024:
    • title: Bayesian Neural Network C++ Library: Review, speaker: C. Juszczak
  • 30.11.2023:
    • title: Hopfield model - neural network based associative memory, speaker: B. Domański
  • 09.11.2023:
    • title: AI Feynman: A Physics-Inspired Method for Symbolic Regression, speaker: R. Durka
  • 26.10.2023:
    • title: Physics Informed Neural Networks, speaker: K. M. Graczyk