Skip to main content
. Author manuscript; available in PMC: 2024 May 15.
Published in final edited form as: Med Image Comput Comput Assist Interv. 2023 Oct 1;14220:651–662. doi: 10.1007/978-3-031-43907-0_62

Fig. 2.

Fig. 2.

Ark-5 and Ark-6 are compared with Google CXR-FM via linear probing (a) with complete training set on six target tasks, demonstrating Ark’s superior or comparable performance and better embedding quality, and (b) with partial training sets or even few-shot samples, showcasing Ark’s outstanding performance in terms of data efficiency.