Friday, May 31, 2013

Visualizations, video/audio, and ML for time series data: which platform?

I want interactive visualizations of a bunch of time series, as well as some audio and video, scrub through them and keep them all synced, and then generate features from them and feed those into a machine learning model. No need to be a web app.

What environment do we do this in? Cross-platform choices seem to be javascript/browser, python/installed, and java/installed.

Visualization:
Javascript: d3 (gallery), cubismGoogle Charts, some others like rickshaw, even processing.js. This is maybe the best reason to pick JS.
Python: matplotlib - mostly static visualizations (gallery) or Bokeh? (see this post) Bokeh outputs to an html5 canvas in the future (or a Chaco plot currently).
Java: ...?

Playing audio/video:
Javascript in a browser: HTML5 video and audio?
Python/Java: beats me. Codec hell?

Machine learning:
Javascript: ...?
Python: scikit-learn (and others)
Java: Weka (I hear the API's a pain, though.)

The path forward seems to be to start building an HTML/JS app, even if it's only client side, and figure something out for the machine learning. Perhaps compile scikit-learn to JS with pyjs? Perhaps (this sounds kind of painful) just send all the features to a server and use weka or scikit-learn there to do the real ML and send back results? But I'd welcome any input.

Saturday, May 25, 2013

It's about how you use it

"I'm still here: back online after a year without the internet"- in which he leaves the internet, and all's well, until it's not anymore; all his bad habits catch up with him offline too. Also, you can't just quit the internet alone. It's part of all our life together. 

Don't be a gargoyle with Google Glass. Don't be with people but not actually with them. Don't tolerate other people being gargoyles.

Is Glass an anti-distraction tool? Could be. So long as you can ruthlessly and efficiently curate what goes into it (or if the algorithms it uses are good at doing that for you). Given people's abilities with their email inboxes, I'm not sure, but I think this writer's on to something: if people have a low cost way to get into your visual field, that'll be a problem.

So what are we left with? "The Internet, or Google Glass, or whatever technology you want, isn't good or bad in itself; it depends on how people use it. (which depends partially on design decisions of the people making it.)" Xkcd is right on.

Three kinds of multitasking

I've got to start posting one thought at a time. Shorter posts more often.

Let's get our words straight: here's multitasking vs. switch-tasking; multitasking is doing a lot of related things, switch-tasking is doing a lot of unrelated things.
Maybe this points to the polychronicity puzzle: how is it that some people can prefer to do many different things, even though multitasking doesn't work? On the other hand, examples of polychronicity sound like switch-tasking, so that actually doesn't help answer that question.
A third term: continuous partial attention, or paying attention to many things at a surface level.

I guess it's like this?


Friday, May 10, 2013

Thinking about time and speed

Half formed ideas and links here.

Been reading a lot of blogs and listening to talks, going to a workshop, thinking about how we view time and how fast it goes. In Motion has a bunch of thoughts about time, in relation to travel; while a lot of the book is frustrating (what is "Deep Travel"? is it just "traveling while paying attention"? why does it need a name?), it makes me think about time as experienced vs. time on the clock.

Monochronic vs. polychronic time - first pointed out to me in two papers by Gloria Mark and Laura Dabbish. Monochronic is what we're used to: one thing at a time, time matters, be on time. Polychronic time sense is the approximate kind of time; do a couple things at once, value relationships more than the time. (but multitasking doesn't work! what's going on here?)

Contemplative Computing points me to a few things: an article about speed mentioning Google Glass cutting down the picture taking time from 12 seconds to 1 second, an interview with Linda Stone on time management vs. attention management, a group interview where Neema Moraveji talks of (among other things) "speed bumps" put into typical tasks. We're putting speed bumps into one task and taking them out of others (and indeed, as texts-on-the-Pebble has shown, removing speed bumps is surprisingly awesome). The answer is probably not "speed is good" or "speed bumps are good", but something more subtle. What is it?

Another thing I want to jot down here: calendars are to time as maps are to space. Ponder that.

Tuesday, May 7, 2013

CHI 2013

... was, of course, great. I student-volunteered, which meant I didn't get to see a ton of talks; nevertheless, here are a few things I particularly liked:

Stories of the Smartphone in Everyday Discourse: Conflict, Tension & Instability, by Ellie Harmon and Melissa Mazmanian. They looked at the stories people tell themselves about their mobile phone use; it oscillates between the "integration" story ("get a smart phone, be a super connected techno future hero") and the "dis-integration" story ("unplug, de-stress, get back to the real world"). This causes tension and makes people uncomfortable. Interestingly, these are kind of the only stories being told, and they're overly simplistic.

Indoor weather stations: investigating a ludic approach to environmental HCI through batch prototyping, by Bill Gaver et al. They put little devices in people's homes that would playfully reflect indoor environment conditions (like slight wind, etc); people didn't find them practical or particularly fun, but still felt some attraction to them; "it's like there's a ghost in the house."

Slow Design for Meaningful Interactions, by Barbara Grosse-Hering et al. They described principles they used to design a juicer. Most interesting: it's okay to make some things slower. It can be good, in fact, as long as they're the key parts of the process. People don't mind spending a little more time juicing; they mind spending a long time cleaning the thing afterwards.

Some cool gadgets you can wear: WatchIt (watch band interaction) and NailDisplay (goes on your fingernail; I wasn't sold at first, but then it grew on me throughout the presentation).


Finally, and most entertainingly, don't use seven segment displays. See you all next year in Toronto!