Circuitry, robot, navigation device, method and network
OP GFlowNets address the inefficiencies in existing blackbox optimization techniques by determining the order of candidate objects within a set of objects and sampling them based on this order, enhancing the efficiency of solving complex optimization problems.
WO2025061851A1 Publication Date: 2025-03-27SONY GROUP CORP +1
0 Cites 0 Cited by
Patent Information
- Application Number
- PCT/EP2024/076276
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-07-09
- Filing Date
- 2024-09-19
- Publication Date
- 2025-03-27
AI Technical Summary
Technical Problem
Existing techniques for solving blackbox optimization problems, which involve optimizing unknown objective functions, are limited in efficiency and require explicit formulations or manual tuning of parameters.
Method used
The use of an order-preserving generative flow network (OP GFlowNet) that determines the order of candidate objects based on an objective function and samples them accordingly, eliminating the need for explicit reward formulations and manual parameter tuning.
Benefits of technology
OP GFlowNets efficiently sample candidate objects that maximize the objective function, improving sample efficiency and handling complex optimization problems like neural architecture search and molecular design.
✦ Generated by Eureka AI based on patent content.
Abstract
The circuitry is configured to generate a candidate object using an order-preserving generative flow network whose input includes a set of objects. The circuitry is configured to determine the order of the candidate object within the set of objects and sample the candidate object based on its determined order.
Need to check novelty before this filing date? Find Prior Art