Speaker
Lévana Gesson
Description
The DINo (Deep learning Intelligence for Nuclear reactiOns) algorithm is a deep neural network designed to improve predictions of nuclear reaction cross-sections, crucial for applications like particle therapy in cancer treatment. Trained on TENDL 2021 data, DINo significantly outperforms traditional models, especially for proton–carbon interactions, achieving better agreement with experimental data. It is efficient, delivering predictions within microseconds, and demonstrates strong generalization, even in data-scarce energy ranges. DINo holds promise for real-time applications and future extension to a broader range of nuclear reactions.
