Print Email Facebook Twitter Finding the Needle in the Pre-Trained Model Zoo Title Finding the Needle in the Pre-Trained Model Zoo: The Use of Rich Metadata and Graph Learning to Estimate Task Transferability Author van der Wilk, Hilco (TU Delft Electrical Engineering, Mathematics and Computer Science) Contributor Hai, R. (mentor) Li, Z. (mentor) Anand, A. (graduation committee) Song, Q. (graduation committee) Degree granting institution Delft University of Technology Programme Computer Science Date 2024-06-25 Abstract The democratization of machine learning through public repositories, often known as model zoos, has significantly increased the availability of pre-trained models for practitioners. However, this abundance can make it difficult to choose the most suitable pre-trained model for fine-tuning on new tasks. Although various methods have been proposed in the field of transferability estimation to address this issue, these methods can take hours to execute and may still fail to find the optimal pre-trained model for fine-tuning. By exploring a new graph learning-based approach to transferability estimation, we outperform state-of-the-art methods such as LogME, improving the accuracy of the best-predicted model by up to 31.5\% in less than 5 minutes. Subject transferability estimationtransfer learningdeep learningmodel zoosgraph learning To reference this document use: http://resolver.tudelft.nl/uuid:59a7191d-4522-4a74-8ccf-503d11a5101b Part of collection Student theses Document type master thesis Rights © 2024 Hilco van der Wilk Files PDF thesis_HvdW_final.pdf 11.54 MB Close viewer /islandora/object/uuid:59a7191d-4522-4a74-8ccf-503d11a5101b/datastream/OBJ/view