Page 264 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications
P. 264
Chapter 9 Applications of deep learning in biomedical engineering 255
16. Fine-tuning stage
It is the supervised learning stage and executes in a top-down
approach. It is mainly responsible for fine-tuning the network pa-
rameters and backpropagates if error occurs [15].
DBN achieves maximum likelihood by using the greedy strat-
egy in each layer [4].
The schema of DBN is shown in Fig. 9.7.
17. Applications of deep learning in
biomedicine
The three major orientations of DL in biomedical applications
can be structured as follows:
(1) DL employs computer-aided diagnosis for efficient diag-
nosing of diseases, so that the burden of physicians will be
eased in the case of a large number of data.
(2) Improved personalized therapies help to improve the medical
care of patients.
(3) To monitor and track the human health, his or her details will
be stored for further analysis, for example, analyzing the
spread of disease and social behaviors with respect to envi-
ronmental factors, or implementation of brainemachine
interface to control wheelchair.
Figure 9.7 Deep belief network.
