testcase

Hi - I have been working on a machine learning library for a few months. While there are a number of ML packages for Max, I wanted one that could be open source (MIT License?), would not add data types of new concepts to max, would use the min-api, and was not media specific. The objects I have made use jitter matrices as data structures because it is easy to convert most things in Max to a matrix. The whole thing is built using the mlpack library (mlpack.org). To be clear, the fact that this uses jitter matrices does not mean it is suitable or usable for things like StyleGAN or pix2pix type of algorithms. The mlpack library is cpu only. It would be cool to see something like tensorflow light integration into Max (i have even been looking into that) but that is not what this is. This package will not contain any type of feature extraction. Basically if whatever you are doing you can get into a jitter matrix these objects will work.

My goal is to get this thing into the package manager but right now I only have it compiled for Mac Intel and Arm. Roughly alpha level quality here and there are a couple known bugs that need to be fixed. Lots of stuff I would change in the code and refactor as well. I have a pull request in for some changes to make auto generation of documentation a bit better but besides that things are working using min-api, mlpack and a couple other libraries needed for mlpack.

What I need help with at this point is compiling on Windows. I would love help from anyone with good C++ skills , people looking to make examples, test things out, familiarity with ML, and people who are fluent in github actions. I will keep plugging away at things but am anxious to see this get out soon. I have linked here to a video which shows how things are working. There is no voice over but some audio examples are in the video so watch you volume. https://vimeo.com/689077561

If interested please comment here or send an email to testcase@asu.edu

Hi - I have been working on a machine learning library for a few months. While there are a number of ML packages for Max, I wanted one that could be open source (MIT License?), would not add data types of new concepts to max, would use the min-api, and was not media specific. The objects I have made use jitter matrices as data structures because it is easy to convert most things in Max to a matrix. The whole thing is built using the mlpack library (mlpack.org). To be clear, the fact that this uses jitter matrices does not mean it is suitable or usable for things like StyleGAN or pix2pix type of algorithms. The mlpack library is cpu only. It would be cool to see something like tensorflow light integration into Max (i have even been looking into that) but that is not what this is. This package will not contain any type of feature extraction. Basically if whatever you are doing you can get into a jitter matrix these objects will work. 

My goal is to get this thing into the package manager but right now I only have it compiled for Mac Intel and Arm. Roughly alpha level quality here and there are a couple known bugs that need to be fixed. Lots of stuff I would change in the code and refactor as well. I have a pull request in for some changes to make auto generation of documentation a bit better but besides that things are working using min-api, mlpack and a couple other libraries needed for mlpack. 

What I need help with at this point is compiling on Windows. I would love help from anyone with good C++ skills , people looking to make examples, test things out, familiarity with ML, and people who are fluent in github actions. I will keep plugging away at things but am anxious to see this get out soon. I have linked here to a video which shows how things are working. There is no voice over but some audio examples are in the video so watch you volume.  https://vimeo.com/689077561

Call for collaborators
Hi - I have been working on a machine learning library for a few months. While there are a number of ML packages for 

call-for-collaborators-for-ml-package

still just learning C++(probably unable to help), but just wanted to praise the effort/project here, and say i'm thoroughly impressed by the video. hope you find the help/support you need.

Thanks for the feedback R∆J∆ - I hope to release something soonish one way or another.

Thanks for the feedback R∆J∆  - I hope to release something soonish one way or another. 

finally got a chance to look at the video, what a fantastic project! congrats on getting to this stage!

in case anyone missed the video link, you should definitely give this a look:

mlmat examples

in case anyone missed the video link, you should definitely give this a look:


call for collaborators for ML package