Abstract
Deep neural networks typically require large amounts of annotated data to be trained effectively. However, in several scientific disciplines, including medical image analysis, generating such large annotated datasets requires specialized domain knowledge, and hence is usually very expensive. In this work, we present a novel application of active learning to data sample selection for training Convolutional Neural Networks (CNN) for Cancerous Tissue Recognition (CTR). Our main idea is to steer annotation efforts towards selecting the most informative samples for training the CNN. To quantify informativeness, we explore three choices based on discrete entropy, best-vs-second-best, and k-nearest neighbor agreement. Our results on three different types of cancer datasets consistently demonstrate that under limited annotated samples, our proposed training scheme converges faster than classical randomized stochastic gradient descent, while achieving the same (or sometimes superior) classification accuracy.
| Original language | English (US) |
|---|---|
| Title of host publication | 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings |
| Publisher | IEEE Computer Society |
| Pages | 1367-1371 |
| Number of pages | 5 |
| ISBN (Electronic) | 9781509021758 |
| DOIs | |
| State | Published - Feb 20 2018 |
| Event | 24th IEEE International Conference on Image Processing, ICIP 2017 - Beijing, China Duration: Sep 17 2017 → Sep 20 2017 |
Publication series
| Name | Proceedings - International Conference on Image Processing, ICIP |
|---|---|
| Volume | 2017-September |
| ISSN (Print) | 1522-4880 |
Other
| Other | 24th IEEE International Conference on Image Processing, ICIP 2017 |
|---|---|
| Country | China |
| City | Beijing |
| Period | 9/17/17 → 9/20/17 |
Fingerprint
Keywords
- Active learning
- Cancer detection
- Deep learning
- Uncertainty sampling
Cite this
Active convolutional neural networks for cancerous tissue recognition. / Stanitsas, Panagiotis; Cherian, Anoop; Truskinovsky, Alexander; Morellas, Vassilios; Papanikolopoulos, Nikolaos.
2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings. IEEE Computer Society, 2018. p. 1367-1371 (Proceedings - International Conference on Image Processing, ICIP; Vol. 2017-September).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Active convolutional neural networks for cancerous tissue recognition
AU - Stanitsas, Panagiotis
AU - Cherian, Anoop
AU - Truskinovsky, Alexander
AU - Morellas, Vassilios
AU - Papanikolopoulos, Nikolaos
PY - 2018/2/20
Y1 - 2018/2/20
N2 - Deep neural networks typically require large amounts of annotated data to be trained effectively. However, in several scientific disciplines, including medical image analysis, generating such large annotated datasets requires specialized domain knowledge, and hence is usually very expensive. In this work, we present a novel application of active learning to data sample selection for training Convolutional Neural Networks (CNN) for Cancerous Tissue Recognition (CTR). Our main idea is to steer annotation efforts towards selecting the most informative samples for training the CNN. To quantify informativeness, we explore three choices based on discrete entropy, best-vs-second-best, and k-nearest neighbor agreement. Our results on three different types of cancer datasets consistently demonstrate that under limited annotated samples, our proposed training scheme converges faster than classical randomized stochastic gradient descent, while achieving the same (or sometimes superior) classification accuracy.
AB - Deep neural networks typically require large amounts of annotated data to be trained effectively. However, in several scientific disciplines, including medical image analysis, generating such large annotated datasets requires specialized domain knowledge, and hence is usually very expensive. In this work, we present a novel application of active learning to data sample selection for training Convolutional Neural Networks (CNN) for Cancerous Tissue Recognition (CTR). Our main idea is to steer annotation efforts towards selecting the most informative samples for training the CNN. To quantify informativeness, we explore three choices based on discrete entropy, best-vs-second-best, and k-nearest neighbor agreement. Our results on three different types of cancer datasets consistently demonstrate that under limited annotated samples, our proposed training scheme converges faster than classical randomized stochastic gradient descent, while achieving the same (or sometimes superior) classification accuracy.
KW - Active learning
KW - Cancer detection
KW - Deep learning
KW - Uncertainty sampling
UR - http://www.scopus.com/inward/record.url?scp=85045310507&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85045310507&partnerID=8YFLogxK
U2 - 10.1109/ICIP.2017.8296505
DO - 10.1109/ICIP.2017.8296505
M3 - Conference contribution
T3 - Proceedings - International Conference on Image Processing, ICIP
SP - 1367
EP - 1371
BT - 2017 IEEE International Conference on Image Processing, ICIP 2017 - Proceedings
PB - IEEE Computer Society
ER -