Page 95 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

Chapter 3 Application, algorithm, tools directly related to deep learning 83

4. Automatic machine translation
Convolution neural networks are useful for identification of
images with visible letters. Once identified, they can be converted
into text, translated, and recreated with an image by using the
translated text. This process is normally called instant visual
translation [25].
5. Automatic handwriting generation
This type generates a new set of handwritings for a given word
or phrase. The handwriting is essentially generated as a sequence
of coordinates used by a pen when the user created samples [24].
The cordial relationship between the pen movement and the let-
ters is learned, and new examples are generated.
6. Imageelanguage translations
A fantastic application of deep Learning is the imagee
language translations. With the Google Translate app, it is now
possible to automatically translate all photographic images with
words into a real-time language of user choice [24]. This is an
extremely useful application and allowing universal humane
world communication [26,27].

5. Conclusion

One of the greatest advantages of using DL approach is its
capability to enact feature engineering using a variety of tools.
In this approach, DL algorithm scans the data to identify features
that correlate and then combine them to promote faster learning
without being told to do so explicitly. This ability comforts data
scientists to save a significant amount of work.

References

[1] https://www.tutorialspoint.com/tensorflow/index.htm.
[2] L. Hao, S. Liang, J. Ye, Z. Xu, TensorD: a tensor decomposition library in
TensorFlow, Neurocomputing 318 (November 27, 2018) 196e200. Elsevier.
[3] https://en.wikipedia.org/wiki/Keras.
[4] N.K. Manaswi, Understanding and working with Keras, in: Deep Learning
with Applications Using Python, @ Springer, Berkeley, CA, 2018, pp. 31e43.
Apress.
[5] https://www.guru99.com/keras-tutorial.html.
[6] D. Kumanov, L.-H. Hung, W. Lloyd, K.Y. Yeung, Serverless Computing
Provides On-Demand High Performance Computing for Biomedical
Research, arXiv preprint arXiv:1807.11659, 2018.
[7] Y. Jia, E. Shelhamer, J. Donahue, S. Karayev, J. Long, R. Girshick,
S. Guadarrama, T. Darrell, Caffe: Convolutional Architecture for Fast
Feature Embedding. Proceedings of the 22nd ACM International
Conference on Multimedia, ACM Digital library, 2014, pp. 675e678.
[8] B.J. Erickson, P. Korfiatis, Z. Akkus, T. Kline, K. Philbrick, Toolkits and
libraries for deep learning, J. Digit. Imaging 30 (2017) 400e405 (@ Springer).