Table 1 Acceleration factor (AF) of multi-fidelity agents in simple acquisitions.

From: Agents for sequential learning using multiple-fidelity data

Agents

Single-fidelity experiments performed for 50% discovery

Multi-fidelity AF\(_{50\%}\)

Single-fidelity experiments performed for 90% discovery

Multi-fidelity AF\(_{90\%}\)

Support vector regression

800

160

1380

160

Random forest regression

740

80

1360

60

GPR\(_{LCB}\)

820

180

1360

220

  1. The AFs are the reduction in number of experiments performed by multi-fidelity agents to achieve a certain amount of discoveries. For each row, we highlighted the agents used, the experiments performed by single-fidelity agents to achieve 50% and 80% discovery, and the acceleration factor of the multi-fidelity agents at those discoveries.
Back to article page