21–30 of 109 posts
Is there any concern about a web-native journal being less "future-proof"? I've come across quite a few interactive learning demonstrations in Flash/Java that no longer work.
I sure hope this catches on, but we should all be aware of the hurdles: - Little incentive for researchers to do this beyond their own good will. - Most ML researchers are bad writers, and it's unlikely that the editing team will do the work needed (which is often a larger reorganization of a paper and ideas) to improve clarity. - Producing great writing and clear, interactive figures, and managing an ongoing github…
In the short term, Distill's editorial assistance will help authors produce outstanding papers, although they need to be willing to work as well.
In the longer-term, I'd like to explore match making between data visualization people who would like to get into machine learning and machine learning researchers publishing papers.
And in the very long term, I think the right solution is to add a new component to the research ecosystem. Just like we we have people who specialize as research engineers, theoreticians, and experimentalists, I'd like to have a respected "research distiller" specialization. Eventually, I'd like to try and start special grants for research groups to have someone focused on this.
I sure hope this catches on, but we should all be aware of the hurdles: - Little incentive for researchers to do this beyond their own good will. - Most ML researchers are bad writers, and it's unlikely that the editing team will do the work needed (which is often a larger reorganization of a paper and ideas) to improve clarity. - Producing great writing and clear, interactive figures, and managing an ongoing github…
Various announcements: Google Research: https://research.googleblog.com/2017/03/distill-supporting-c... DeepMind: https://deepmind.com/blog/distill-communicating-science-mach... OpenAI: https://openai.com/blog/Distill/ YC Research: http://blog.ycombinator.com/distill-an-interactive-visual-jo... Chris Olah: http://colah.github.io/posts/2017-03-Distill/
Awesome. How do I donate to this?
I worked in clinical research in a past life and studies would be highly discounted if they couldn't be reproduced. A highly detailed methods section was key. Many ML papers I see tend to have incredibly formalized LaTeX+Greek obsessed methods section, but far short of anything to allow reproduction. Some ML papers, i swear must have run their parameter searches a 1000 times to overfit and magically achieve 99% AUC.
Worse, I actually have tons of spare GPU farm capacity i'd love to devote to re-producing research, tweaking, trying it on adjacent datasets, etc. But the effort to re-produce is too high for most papers.
It is also disappointing to see various input datasets strewn about individuals' personal homepages, and sometimes end up broken. Sometimes the "original" dataset is in a pickled form after having already gone through multiple upstream transformations. I hope Distill can instill some good best practices to the community.
I don't want to undermine visualizations, they are awesome, but one of the big problems I see with ML research is the lack of re-produceability. I know that Google, Facebook and some others already share associated source repos, but it should almost be mandatory when working with public benchmark datasets. Source + Docker Images would be even better. I worked in clinical research in a past life and studies would be h…
It seems clear to me that the future will involve some kind of linking reproducibility to papers. If we want to find that future, we need a way for people to experiment with what a publication is.
As a developer with a weaker background in mathematics, I face a language barrier with many modern algorithms. After lots of research I can understand and explain them in code, but I have no idea what your artistic-looking MathXML means. Visualizations or algorithms described using code are much, much easier for me to understand and serve as a great starting point for unpacking the math explanations.
I would recommend picking up a book on Comp Sci or algorithms, even just a cursory reading helps a lot. CS is very much not just programming and it is heavily restricted by descriptions through code.
I sure hope this catches on, but we should all be aware of the hurdles: - Little incentive for researchers to do this beyond their own good will. - Most ML researchers are bad writers, and it's unlikely that the editing team will do the work needed (which is often a larger reorganization of a paper and ideas) to improve clarity. - Producing great writing and clear, interactive figures, and managing an ongoing github…
I clearly don't expect people to do that much. I can only do that because i'm coming from web development, and very nice tools started to appear recently.
People in research needs a design framework like a set of templates for keynotes/PPT/JS/CSS (think about how much traction got bootstrap). Distill is doing an awesome jobs at showing the example of what you could do.
Maybe Distill could open-source the templates they use to build those blog post?
The announcements and About page indicate an emphasis on visuals and presentation, which I apprI've. But when I think of "modern machine learning," I think of open-source and reproducibility (e.g. Jupyter notebooks). Will the papers published on Distill maintain transparency of the statistical process? I see in the submission notes that articles are required to be a public GitHub repo, which is a positive indicator.…