Verstärkungslernprogramm, Verfahren zum Verstärkungslernen und Informationsverarbeitungsvorrichtung

Details

Veröffentlichungs-Nr.

EP4734119

Anmeldetag

23. Oktober 2025

Veröffentlichung

29. April 2026

Rechtsraum

EP

IPC

G16C20/50, G16C20/70(KUMAR NEERAJ et al.)(ISHITANI RYUICHIRO et al.)(UNIV CALIFORNIA et al.)

Abstract

A reinforcement learning program that causes a computer to execute a process including: executing reinforcement learning of a machine learning model (21a), using a first three-dimensional (3D) molecular structure of a first compound to be adapted to a first object as a state, assignment of a type of an additional atom, a binding position of the additional atom to the first 3D molecular structure, and a 3D position of the additional atom, to the first compound in the state as an action, and an index to evaluate a second 3D molecular structure obtained by binding the additional atom to the first 3D molecular structure in the action as a reward.

Anmelder

Firma

FUJITSU LIMITED

Land

🇯🇵 Japan

Fachgebiete