Dictionary of Applied Machine Learning
Updated on 2026-09-18
Typeset PDF version — the authoritative form of this entry
Transfer learning is the reuse of what was learned on one learning task to solve another learning task, typically one with fewer labeled data points. A common form starts from a deep net trained on a large dataset, keeps the features computed by its early layers, and adapts the remaining model parameters to the new task by fine-tuning on the small dataset. The two tasks may differ in their probability distribution of data points, in their labels, or in both; transfer helps when the features useful for the first task remain useful for the second. For example, a deep net trained on natural images is fine-tuned on a small medical imaging dataset to detect tumors.
Transfer learning aims at leveraging information obtained while solving an existing learning task to solve another learning task. For example, a deep net trained on a large natural-image dataset such as ImageNet can be fine-tuned on a smaller medical imaging dataset to detect tumors, transferring learned visual features to the new task.
See also: learning task, multitask learning.
@misc{dictml_transferlearning,
author = {Jung, Alexander and Olioumtsevits, Konstantina and Schnoor, Ekkehard},
title = {transfer learning},
howpublished = {Dictionary of Applied Machine Learning (course edition)},
year = {2026},
doi = {10.5281/zenodo.21569296},
note = {ISBN 978-952-64-3013-3, CC BY 4.0, retrieved 2026-09-19},
url = {https://dictionaryofml.org/terms/transferlearning.html}
}