{"id":17537,"date":"2018-04-16T12:00:23","date_gmt":"2018-04-16T11:00:23","guid":{"rendered":"https:\/\/blogs.nature.com\/naturejobs\/?p=17537"},"modified":"2018-05-08T19:04:42","modified_gmt":"2018-05-08T18:04:42","slug":"machine-learning-gets-a-journal-for-interactive-figures","status":"publish","type":"post","link":"https:\/\/blogs.nature.com\/naturejobs\/2018\/04\/16\/machine-learning-gets-a-journal-for-interactive-figures\/","title":{"rendered":"Machine learning gets a journal for interactive figures"},"content":{"rendered":"

\"Screen<\/a><\/p>\n

Distill wants to be a sandbox for what a scientific paper can be <\/em><\/p>\n

By Anna Nowogrodzki <\/strong><\/p>\n

Sometimes it\u2019s hard to understand someone else\u2019s research through a static scientific paper. Across countless universities and companies, at whiteboards and cafeteria tables, you\u2019ll find scientists in animated conversations explaining their research to one another, asking questions, playing around with each other\u2019s data: in short, interacting. Across the internet in recent years, people have extended these explanations to include interactive graphics and code.<\/p>\n

Now a web-only machine-learning journal called Distill<\/em><\/a> aims to provide a formal home for these interactive graphical explanations, which in recent years have expanded to blogs and other online fora.<\/p>\n

<\/p>\n

\u201cI think what they\u2019re doing is really nice, and it’s kind of formalizing something that\u2019s already happening informally,\u201d says Jennifer Listgarten<\/a>, a computational biologist at University of California at Berkeley. \u201cPeople are posting blogs, or on GitHub repositories, IPython notebooks\u2014things where you can actually really get a more intuitive feel than you can from just the traditional PDF.\u201d Now the journal, which launched in March 2017, is making it possible to publish these communications directly by providing detailed design mentorship and custom web components that make building interactive papers easier than with raw HTML, CSS or JavaScript.<\/p>\n

Distill\u2019s editorial team, led by Google research scientists and co-founders Christopher<\/a> Olah<\/a> and Shan Carter<\/a>, believes that interactive images and figures can often communicate better than static ones. Example figures have shown how a neural network decides whether an image is of a dog or a cat<\/a>, and how momentum works<\/a>. Readers can mouse over parts of a dog image to see trippy visualizations of what individual neurons in the network \u201csee,\u201d or drag slider bars adjusting step size and other variables to show how momentum creates its own oscillations.<\/p>\n

By publishing these figures in an academic journal, Distill<\/em> hopes to help researchers gain recognition for the work they do in creating them. \u201cWe saw over and over again people doing this kind of work and it not being able to help their academic careers,\u201d says Olah.<\/p>\n

The journal also aims to boost reproducibility by making backend code available so that readers can more quickly and easily replicate results. For Olah\u2019s recent article about how machine learning “sees” and interprets photos<\/a>, nearly 2,500 of the article\u2019s 93,351 readers (2.6%) have interacted with the associated code notebook<\/a>. That notebook allows readers to recreate the interactive graphics and reproduce the contributing analyses , says Olah, a feature that is available by simply clicking on links under each graphic that say \u201creproduce in a notebook.\u201d \u201cThe paper is sort of under this permanent and ongoing review by the public,\u201d Olah explains.<\/p>\n

As technology editor Jeffrey Perkel recently wrote\u00a0in\u00a0Nature<\/em> Toolbox, several other publications and publishers\u00a0also support interactive graphics and code<\/a>, including F1000Research<\/em>, GigaScience<\/em>, IEEE, and SPIE. But for Distill,<\/em> dynamic graphics are not just supported but an area of primary focus. That approach is especially suited to machine learning, says Listgarten. \u201cIt\u2019s very easy, especially in machine learning, for people to just \u2026 throw down a lot of equations and try to wow people with their technical prowess,\u201d she says. \u201cBut at the end of the day, the best style of communication is often intuitive and pictorial.\u201d<\/p>\n

So far, only 10 papers have been published on Distill;<\/em> all have been assigned DOIs. \u201cWe\u2019re not publishing that many papers. We\u2019re trying to create a prototype,\u201d says Olah, of \u201ca particular way of communicating scientific research. A lot of our papers are pushing in a new direction for what a paper can be.\u201d<\/p>\n

It\u2019s not easy. Supporting dynamic figures involves intensive web development and design thinking, says Olah. Authors create the figures using tools such as the JavaScript D3.js library (for interactive figures) and Adobe Illustrator, Sketch or Inkscape (for static figures). Review of a paper includes not just traditional review elements, but also rigorous design feedback from Olah and Carter. The entire review process is documented in each paper\u2019s GitHub repository<\/a>, which is made public when the paper is published.<\/p>\n

Olah, who has authored six\u00a0papers in Distill<\/em>, says the process is different from \u2013 and requires a greater time commitment than — traditional research publishing. \u201cI\u2019m investing so much more energy into it, sharing this way of thinking instead of just listing a bunch of results. It\u2019s also a lot more work\u2014I wouldn\u2019t want to fool anyone. It\u2019s a different kind of labor, and it\u2019s as difficult as doing research.\u201d<\/p>\n

But it\u2019s worth the effort, he adds. \u201cI really like it when people are like, \u2018Reading this was a pleasure.\u2019 It feels like I got to make somebody happy.\u201d And that\u2019s not your typical scientific publishing experience.<\/p>\n

Anna Nowogrodzki is a science writer based in Boston, Massachusetts<\/em>.<\/p>\n

 <\/p>\n

Image credit<\/strong>: Goh, G. Why momentum really works. Distill<\/em> (2017). https:\/\/doi.org\/10.23915\/distill.00006<\/a>. Licensed under CC-BY 2.0<\/a>.<\/p>\n

 <\/p>\n

Correction<\/strong> (8 May 2018): Jennifer Listgarten’s name was originally misspelled. <\/em>Nature regrets the error.\u00a0<\/em><\/p>\n

 <\/p>\n

Suggested posts<\/strong><\/p>\n

Software quality tests yield best practices<\/a><\/p>\n

‘Manubot’ powers a crowdsourced ‘deep-learning’ review<\/a><\/p>\n

eLife replaces commenting system with Hypothesis annotations<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Distill wants to be a sandbox for what a scientific paper can be  … Read more<\/a> Continue reading →<\/span><\/a><\/p>\n","protected":false},"author":104777,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[192,359,200],"tags":[10985385,10985387,4827723,10139713,6677357,75],"class_list":["post-17537","post","type-post","status-publish","format-standard","hentry","category-blog-2","category-publishing-2","category-technology-2","tag-anna-nowogrodzki","tag-christopher-olah","tag-distill","tag-interactive-figures","tag-techblog","tag-technology"],"_links":{"self":[{"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/posts\/17537","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/users\/104777"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/comments?post=17537"}],"version-history":[{"count":0,"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/posts\/17537\/revisions"}],"wp:attachment":[{"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/media?parent=17537"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/categories?post=17537"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.nature.com\/naturejobs\/wp-json\/wp\/v2\/tags?post=17537"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}