Home > On-Demand Archives > Q&A Sessions >

Live Q&A - Building A Tensorflow Lite Neural Network Vibration Classifier, With A Little Help From DSP

John Edwards - Watch Now - DSP Online Conference 2022 - Duration: 22:54

Live Q&A - Building A Tensorflow Lite Neural Network Vibration Classifier, With A Little Help From DSP
John Edwards
M↓ MARKDOWN HELP
italicssurround text with
*asterisks*
boldsurround text with
**two asterisks**
hyperlink
[hyperlink](https://example.com)
or just a bare URL
code
surround text with
`backticks`
strikethroughsurround text with
~~two tilde characters~~
quote
prefix with
>

Kelly_C
Score: 0 | 2 years ago | 1 reply

Hi John, nice talk. Where would I find tutorial background information that would help with design tradeoffs?

john.edwardsSpeaker
Score: 0 | 2 years ago | no reply

Hi Kelly,
Thank you very much for your kind words.
Unfortunately, I'm not aware of any material of that kind, I guess because it is such a new topic of interest.
The best thing to do is search for published articles. This is a good example, although it uses Wavelets rather than the FFT - https://www.hindawi.com/journals/sv/2020/1650270/
Good luck in your search.
Best regards,
John

john.edwardsSpeaker
Score: 0 | 2 years ago | no reply

Please do submit any questions here and I will be glad to answer them.
I'll be online straight after the video finishes
Source code and test datasets can be downloaded from here: https://github.com/Numerix-DSP/DSP_And_ML_Examples

10:31:53	 From  Leonard : hello
10:32:17	 From  John Edwards : Hi Everyone, Welcome to the live Q&A
10:46:22	 From  Michael Kirkhart : Here is a URL for a potential dataset for motor vibration classification: http://www02.smt.ufrj.br/~offshore/mfs/page_01.html
10:47:32	 From  John Edwards : Thanks Michael, yes that's another one
10:53:13	 From  Michael Kirkhart : Great talk - excellent application example - like the fact that you did not blindly throw the data at the CNN, but employed "old school" feature engineering.
10:54:05	 From  Michael Kirkhart : As always, it depends

OUR PARTNERS