SKM 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 22: Machine Learning in Dynamics and Statistical Physics II
DY 22.2: Vortrag
Dienstag, 28. März 2023, 14:15–14:30, ZEU 160
Studying sequence property relationships with neural networks — •Huzaifa Shabbir1, Jens Uwe Sommer1,2, and Marco Werner1 — 1Leibniz Institute for Polymer Research Dresden, Germany. — 2Technische Universität Dresden
In this work, we investigate the relationships between chemical sequence and property space for various sequence lengths with the help of neural networks (NN). Two different systems are investigated for this purpose: system I comprises copolymer sequences and their free energy of interaction with a lipid bilayer membrane. System II consists of metallic nanoparticle sequences and their plasmonic spectrum. We compare the performance of different neural network architectures such as feed-forward NNs and gated recurrent unit (GRU) networks in terms of their interpolation and extrapolation capacity between different sequence lengths. We show that the GRU is particularly suitable to transfer the learned patterns from smaller sequence lengths to enhance significantly the learning result for larger sequence lengths.