Is it possible to train numeric data using transfer learning?
4 views (last 30 days)
Show older comments
Byung - Gi Jeon
on 20 Sep 2020
Commented: Byung - Gi Jeon
on 21 Sep 2020
I did research on learning large amounts of numeric data with RNN (LSTM).
The results were pretty accurate.
However, recently, I would like to compare result CNN and transfer learning.
I have studied many matlab examples documents and search on google.
However, I couldn't find an example of learning 'double(numeric)' data. Everything is for images.
So my question is, is it possible learn double data using transfer learning? If so, can you give me an example or material I can refer to?
I really sorry for the worst English skills.
Thank you.
0 Comments
Accepted Answer
Walter Roberson
on 20 Sep 2020
Yes, it is possible to use transfer learning for non-image data.
For example https://becominghuman.ai/unsupervised-pre-training-for-speech-recognition-wav2vec-aba643824324?gi=f24550e74b12 specifically says that it can be done, and has a link to python code for the purpose.
The concepts are the same as for image recognition; the problem is finding a pre-trained network that can be usefully adapted for your purposes.
More Answers (0)
See Also
Categories
Find more on Image Data Workflows in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!