Lifelong deep learning aplicada na melhoria da classificação de imagens de câncer de mama
| Main Authors: | Cotta, Paulo V. P., da Silva, Nilton Correia, Jesus, Geovanni, Gomes, Marília Miranda Forte |
|---|---|
| Format: | Proceeding |
| Bahasa: | por |
| Terbitan: |
, 2019
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/3457582 |
Daftar Isi:
- This paper proposes the creation of a continuous learning process for deep neural networks, aiming to identify medical images, focusing on breast cancer. In the Lifelong Learning Continuous Learning Process you can identify if your learning has been decaying, and if so, the Lifelong Deep Learning technique will perform the learning process again or substitute for improved learning. The purpose of the research is to use deep networks to classify breast cancer and apply Lifelong Deep Learning to the continuous learning process. Deep convolutional networks and the knowledge transfer process with reinforcement learning will be detailed. Where together with the two techniques a rating value of 63% was achieved.
- XII SIMPÓSIO DE ENGENHARIA BIOMÉDICA - IX SIMPÓSIO DE INSTRUMENTAÇÃO E IMAGENS MÉDICAS