Open Call for 2017 Fellowships

REAS

The Processing Foundation is currently accepting applications for the 2017 Fellowship Program.

Application Deadline: December 19, 2016, 11:59PM PST

Processing Foundation Fellowships support artists, coders, and collectives in visionary projects that conceive a new direction for what Processing as a software and a community can do. Fellowships are an integral part of the Processing Foundation’s work toward developing tools of empowerment and access at the convergence of art and technology.

All of the details are on the Foundation website

I'll try to answer questions here as they come up.

kaichogami

As my fellowship project, I want to extend processing library for visualizing machine learning concepts. For practitioners not from technical background, (especially using deep learning), visualizing could be perhaps the easiest way to understand the high level idea. This project could be implemented in several ways, depending on what we are trying to visualise.

1) Visualisation of transformed data to understand what the data is representing. In machine learning preprocessing steps such as PCA(principle component analysis), LDA(Linear discriminant analysis) helps to reduce the feature space and bring out the latent features. What these features represents is essential for understanding the essence of machine learning. This is even more vital in neural network where hidden layers will represent transformation of features in huge dimensions. This is currently a hot topic in study and I believe understanding the representation of lower dimension feature space is starting step for understanding complicated neural networks. As data transformations are being visualized, users will have to write their own code for machine learning algorithms, with functionality to visualize at every step of the algorithm. This however seems not so friendly with people not familiar with technical details.

2) Visualizing several machine learning algorithms. This will be very specific in nature and aimed at people that are beginning to learn about the working of machine learning algorithms. Algorithm could belong to different group of machine learning categories such as clustering, Bayesian, decision tree etc and they behave differently. Here each algorithm will have its own visualization which will vary by changing the algorithm parameters. This will be friendly with non technical background practitioners.

3) We could possible combine and visualize both transformations as well as algorithms separately for much wider audience. This will however be much larger in scope.

These visualization could be then animated. For example, in linear regression, it will nice to animate the line as it finds the best fit. Being new to machine learning and processing I want to know whether it is a feasible idea and whether it lies within the scope of processing's goal. My question then lies, what could be the best possible approach for creating a library that helps new people to get started with machine learning using nice visualizations?

Thank you

Asish Panda

REAS

Hello Asish, please complete the Fellowship form here: https://docs.google.com/forms/d/e/1FAIpQLSc2amDgi-lHQ1-C_-2goyP1TXNahWqxqvFldXxFDKDav7zxlA/viewform?c=0&w=1

kaichogami

Hello Reas, thanks for the heads up. However I was looking for a discussion what approach would be most beneficial, for machine learning users. Apologies for posting it in the wrong topic.

Thank you

Asish Panda

REAS

OK, this is the correct spot for a discussion. I wasn't sure and I wanted to make sure you knew about the formal submission process.

Akarshit

Hey all, I would like to work on the p5-web-editor during the fellowship. The editor is still in the development phase so there are a lot of issues/features to work on. Some of these interesting to me are:

1. Implementation of find This is a basic feature that is almost a necessity. The implementation would include find/replace.

2. Limiting file upload to S3 This should be high priority otherwise some people would just use this to store their content.

3. Local storage for internet connectivity issues - kind of what google docs does so the editor would give a good experience irrespective of the internet speed and when the connection is good it will sync to the server.

4. Add lib reference This would be very useful to the users of web-editor and its presence is now almost considered a standard in IDEs.

I would be very interested in getting suggestion from the community about the ideas. Or if there is some thing having a higher priority, I would be happy to work on it.

chandni

The Google form said to expect an intimation by Jan 15 if selected. Can somebody confirm if that's happened? I couldn't tell if we'd be emailed if the application was not selected.

blyk

Till now even I have't got any notification so I think only select participants are going to receive the email but it would be great if we get notified.

colouredmirrorball

They received a lot of applications and the announcement will be delayed

https://facebook.com/story.php?story_fbid=10154087095297651&id=49439222650