Comparison of Deep Transfer Learning Strategies for Digital Pathology
By Romain Mormont, Pierre Geurts, Raphaël Marée. 2018
119 citations recorded by the source corpus.
In this paper, we study deep transfer learning as a way of overcoming object recognition challenges encountered in the field of digital pathology. Through several experiments, we investigate various uses of pre-trained neural network architectures and different combination schemes with random forests for feature selection. Our experiments on eight classification datasets show that densely connected and residual networks consistently yield best performances across strategies. It also appears that network fine-tuning and using inner layers features are the best performing strategies, with the former yielding slightly superior results.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Quoc V. Le
- Alan L. Yuille
- Aleksander Madry
- Jiashi Feng
- Percy Liang
- Logan Engstrom
- Jianfeng Gao
- Nicholas Carlini
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Alan L. Yuille
- Ming-Hsuan Yang
- Trevor Darrell
- Bernt Schiele
- Xiaogang Wang
- Jitendra Malik
In Representation & Self-Supervised Learning
- Quoc V. Le
- Simon Kornblith
- Jonathon Shlens
- Stephen Lin
- Aleksander Madry
- Andrew Ilyas
- Logan Engstrom
- Zhirong Wu
Citations in the corpus
- Do Better ImageNet Models Transfer Better? (2018, background): “A substantial body of existing research indicates that, in image tasks, fine-tuning typically achieves higher accuracy than classification based on fixed features, especially for larger datasets or datasets with a larger domain mismatch from the training set [1, 6, 24, 74, 2, 43, 33, 9, 51].”
- Do Adversarially Robust ImageNet Models Transfer Better? (2020, background): “Indeed, transfer learning has been studied in varied domains including medical imaging [MGM18], language modeling [CK18], and various object detection and segmentation related tasks [Ren+15; Dai+16; Hua+17; Che+17].”
- What makes instance discrimination good for transfer learning? (2020, background): “ImageNet pretraining also works amazingly well under a large domain gap for medical imaging [22] and depth estimation [18].”
- A Review of Deep Transfer Learning and Recent Advancements (2022, background): “[31] 2018 Comparison of Deep Transfer Learning Strategies for Digital Pathology Image no Freezing CNN layers yes NA Medical image”
- Rethinking the Hyperparameters for Fine-tuning (2020, unknown): “It has been shown that fine-tuning the whole network usually results in better performance than using the network as a static feature extractor (Yosinski et al., 2014; Donahue et al., 2014; Huh et al., 2016; Mormont et al., 2018; Kornblith et al., 2019).”