Microprocessors

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

Microprocessors

steveinjava .
Hi Weka List! I am a newbie, fresh from a starter course on FutureLearn. Although I am fairly competent in Java I want to use machin learning with microprocessors.
The plan is to have some water bottle catamarans (don't laugh!) with simple sensors controlled by ESP32 with Lora radio links to a pc based on the swimming pool wall. The sensed information will be sent in as data for Weka and I need the resultant improved algorithm to be then sent back to the ESP32's on the boats. Now ESP32 are great but they can't handle even J2ME. It has to be C. Is there anybody working on a C output for Weka? If not, can anybody recommend a different machine learning tool that might? Many thanks in advance :)

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

Peter Reutemann
> Hi Weka List! I am a newbie, fresh from a starter course on FutureLearn. Although I am fairly competent in Java I want to use machin learning with microprocessors.
> The plan is to have some water bottle catamarans (don't laugh!) with simple sensors controlled by ESP32 with Lora radio links to a pc based on the swimming pool wall. The sensed information will be sent in as data for Weka and I need the resultant improved algorithm to be then sent back to the ESP32's on the boats. Now ESP32 are great but they can't handle even J2ME. It has to be C. Is there anybody working on a C output for Weka? If not, can anybody recommend a different machine learning tool that might? Many thanks in advance :)

Some classifiers (like J48) support source code generation from build models:
https://weka.sourceforge.io/doc.dev/weka/classifiers/Sourcable.html

You should be able to convert that code (manually) into C.

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 858-5174
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

佐藤 広行
In reply to this post by steveinjava .

Hello

 

What about obniz, which supports JavaScript, not Java, but better than C.

 

Hiro

 

From: steveinjava . [mailto:[hidden email]]
Sent: Thursday, May 21, 2020 8:26 PM
To: [hidden email]
Subject: [Wekalist] Microprocessors

 

Hi Weka List! I am a newbie, fresh from a starter course on FutureLearn. Although I am fairly competent in Java I want to use machin learning with microprocessors.

The plan is to have some water bottle catamarans (don't laugh!) with simple sensors controlled by ESP32 with Lora radio links to a pc based on the swimming pool wall. The sensed information will be sent in as data for Weka and I need the resultant improved algorithm to be then sent back to the ESP32's on the boats. Now ESP32 are great but they can't handle even J2ME. It has to be C. Is there anybody working on a C output for Weka? If not, can anybody recommend a different machine learning tool that might? Many thanks in advance :)


_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

Peter Reutemann
> What about obniz, which supports JavaScript, not Java, but better than C.

Machine learning library in JS:
https://github.com/mljs/ml

Found through this blog entry:
https://blog.bitsrc.io/11-javascript-machine-learning-libraries-to-use-in-your-app-c49772cca46c

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 858-5174
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

steveinjava .
In reply to this post by Peter Reutemann
Peter, many thanks for the reply. I am new to Weka but am surprised all classifiers do not produce code for a model. For my purpose, I have to have direct code I can upload- there is no place for converting manually as I want autonomous updates of the model as fresh data arrives. I will research further. Thanks again, Steve

On Fri, 22 May 2020 at 00:34, Peter Reutemann <[hidden email]> wrote:
> Hi Weka List! I am a newbie, fresh from a starter course on FutureLearn. Although I am fairly competent in Java I want to use machin learning with microprocessors.
> The plan is to have some water bottle catamarans (don't laugh!) with simple sensors controlled by ESP32 with Lora radio links to a pc based on the swimming pool wall. The sensed information will be sent in as data for Weka and I need the resultant improved algorithm to be then sent back to the ESP32's on the boats. Now ESP32 are great but they can't handle even J2ME. It has to be C. Is there anybody working on a C output for Weka? If not, can anybody recommend a different machine learning tool that might? Many thanks in advance :)

Some classifiers (like J48) support source code generation from build models:
https://weka.sourceforge.io/doc.dev/weka/classifiers/Sourcable.html

You should be able to convert that code (manually) into C.

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 858-5174
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


--
Regards, Steve H

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

steveinjava .
In reply to this post by 佐藤 広行
Hiro, thanks for the info but it has to be code the microprocessor can execute without conversion. I will find what I need eventually. Thanks again, Steve

On Fri, 22 May 2020 at 01:21, 佐藤 広行 <[hidden email]> wrote:

Hello

 

What about obniz, which supports JavaScript, not Java, but better than C.

 

Hiro

 

From: steveinjava . [mailto:[hidden email]]
Sent: Thursday, May 21, 2020 8:26 PM
To: [hidden email]
Subject: [Wekalist] Microprocessors

 

Hi Weka List! I am a newbie, fresh from a starter course on FutureLearn. Although I am fairly competent in Java I want to use machin learning with microprocessors.

The plan is to have some water bottle catamarans (don't laugh!) with simple sensors controlled by ESP32 with Lora radio links to a pc based on the swimming pool wall. The sensed information will be sent in as data for Weka and I need the resultant improved algorithm to be then sent back to the ESP32's on the boats. Now ESP32 are great but they can't handle even J2ME. It has to be C. Is there anybody working on a C output for Weka? If not, can anybody recommend a different machine learning tool that might? Many thanks in advance :)

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


--
Regards, Steve H

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

steveinjava .
In reply to this post by Peter Reutemann
Peter, thank again, I will research all these links. Ultimately, it has to be C so it can be uploaded autonomously. I did, in a former life, write a transpiler between pascal and basic so its not impossible although I will keep looking for a more direct method.
All the best from around the globe, Steve

On Fri, 22 May 2020 at 02:04, Peter Reutemann <[hidden email]> wrote:
> What about obniz, which supports JavaScript, not Java, but better than C.

Machine learning library in JS:
https://github.com/mljs/ml

Found through this blog entry:
https://blog.bitsrc.io/11-javascript-machine-learning-libraries-to-use-in-your-app-c49772cca46c

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 858-5174
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html


--
Regards, Steve H

_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

Peter Reutemann-3
In reply to this post by steveinjava .
On May 23, 2020 2:45:05 AM GMT+12:00, "steveinjava ." <[hidden email]> wrote:

>Peter, many thanks for the reply. I am new to Weka but am surprised all
>classifiers do not produce code for a model. For my purpose, I have to
>have
>direct code I can upload- there is no place for converting manually as
>I
>want autonomous updates of the model as fresh data arrives. I will
>research
>further. Thanks again, Steve
>
>On Fri, 22 May 2020 at 00:34, Peter Reutemann <[hidden email]>
>wrote:
>
>> > Hi Weka List! I am a newbie, fresh from a starter course on
>FutureLearn.
>> Although I am fairly competent in Java I want to use machin learning
>with
>> microprocessors.
>> > The plan is to have some water bottle catamarans (don't laugh!)
>with
>> simple sensors controlled by ESP32 with Lora radio links to a pc
>based on
>> the swimming pool wall. The sensed information will be sent in as
>data for
>> Weka and I need the resultant improved algorithm to be then sent back
>to
>> the ESP32's on the boats. Now ESP32 are great but they can't handle
>even
>> J2ME. It has to be C. Is there anybody working on a C output for
>Weka? If
>> not, can anybody recommend a different machine learning tool that
>might?
>> Many thanks in advance :)
>>
>> Some classifiers (like J48) support source code generation from build
>> models:
>> https://weka.sourceforge.io/doc.dev/weka/classifiers/Sourcable.html
>>
>> You should be able to convert that code (manually) into C.
>>
>> Cheers, Peter
>> --
>> Peter Reutemann
>> Dept. of Computer Science
>> University of Waikato, NZ
>> +64 (7) 858-5174
>> http://www.cms.waikato.ac.nz/~fracpete/
>> http://www.data-mining.co.nz/
>> _______________________________________________
>> Wekalist mailing list -- [hidden email]
>> Send posts to [hidden email]
>> To unsubscribe send an email to [hidden email]
>> To subscribe, unsubscribe, etc., visit
>>
>https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
>> List etiquette:
>> http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
>>

Generating code output (which doesn't rely on any Weka data structures) is extra code that is tedious to write and hard to maintain (think large matrices and matrix operations.
Weka stores models either as Java serialized objects or via XML serialization. This is what most people use. Of course, it requires the Weka library to be present.

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 858-5174
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: Microprocessors

Peter Reutemann-3
In reply to this post by steveinjava .
On May 23, 2020 2:53:03 AM GMT+12:00, "steveinjava ." <[hidden email]> wrote:

>Peter, thank again, I will research all these links. Ultimately, it has
>to
>be C so it can be uploaded autonomously. I did, in a former life, write
>a
>transpiler between pascal and basic so its not impossible although I
>will
>keep looking for a more direct method.
>All the best from around the globe, Steve
>
>On Fri, 22 May 2020 at 02:04, Peter Reutemann <[hidden email]>
>wrote:
>
>> > What about obniz, which supports JavaScript, not Java, but better
>than C.
>>
>> Machine learning library in JS:
>> https://github.com/mljs/ml
>>
>> Found through this blog entry:
>>
>>
>https://blog.bitsrc.io/11-javascript-machine-learning-libraries-to-use-in-your-app-c49772cca46c
>>
>> Cheers, Peter
>> --
>> Peter Reutemann
>> Dept. of Computer Science
>> University of Waikato, NZ
>> +64 (7) 858-5174
>> http://www.cms.waikato.ac.nz/~fracpete/
>> http://www.data-mining.co.nz/
>> _______________________________________________
>> Wekalist mailing list -- [hidden email]
>> Send posts to [hidden email]
>> To unsubscribe send an email to [hidden email]
>> To subscribe, unsubscribe, etc., visit
>>
>https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
>> List etiquette:
>> http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
>>

Have a look here:

https://eloquentarduino.github.io/2020/01/easy-tinyml-on-esp32-and-arduino/

You might want to search for "TinyML" and "edge machine learning".

Cheers, Peter
--
Peter Reutemann
Dept. of Computer Science
University of Waikato, NZ
+64 (7) 858-5174
http://www.cms.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [hidden email]
Send posts to [hidden email]
To unsubscribe send an email to [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html