Engineer on the SageMaker team here.

There are no restrictions on the types of algorithms that you can optimize using the HyperParameterOptimization service.

SageMaker is designed for machine learning which means it's optimized for algorithms that process a lot of data to develop a model where each run of the algorithm may generate an objective function value (or potentially many such as the value may change during training).

If this structure fits your problem, SageMaker could be useful to you even if your problem isn't strictly "machine learning."

That's great. Thank you for following up so quickly. Could you point me to where I could read about this, or see an example?

I took a quick search on the documentation and I could see anything on a cursory pass:

http://docs.aws.amazon.com/search/doc-search.html?searchPath...

Sorry, for the very delayed reply here.

The hyperparameter optimization feature is still in preview (though the rest of SageMaker is in general availability). We'll be putting up a page within the next week or so for you to request access.

Google is a data company and the data available has changed over the last few years. Maps are both a way to present data and a way to collect data.

The new data that Google has comes from Android handsets and from users using Google maps and Waze on Android and non-Android handsets.

This data is all about users in motion. At the scale shown in this article, it's almost exclusively people driving. As a result, it makes a lot more sense to focus on the connections over the places they connect. This becomes clearer when you view the roads as more active entities by including congestion and other real-time data.

This may not be the best presentation for everyone, but it seems to be the presentation that fits best with Google's current mission and capabilities.

Of all of these, #3 has made the most memorable conferences in my career. I feel like I've gotten value when I get on the plane to go home and I can't decide which of 20 new projects I've thought of in the last 3 days that I want to start on first.

Of course, #3 depends heavily on #1 & #2.

Yeah, but you have to clear your browser and reload. The instructions are here: http://deliciousengineering.blogspot.com/ under "Firefox Extension Fixed." They claim that the IE extension is fixed as well, but there seem to be no special instructions.

Unlike environments like GWT or ClojureScript, coffeescript doesn't treat the JavaScript as an invisible target language. You do actually think a little about the JavaScript that your code is being compiled into (or maybe more accurately, translated into).

This makes CoffeeScript fit really well into the JS-centric environment that has grown up around tools like JQuery and backbone. It's also very easy to debug in the browser compared to the environments that are targeting JS as an object code.

Personally, I don't think of CoffeeScript as a fully independent language but, rather, as a way to write JS which is cleaner, more reliable, and more fun.

If you really want to turn JS into a black box (and there are some good arguments for that), GWT or ClojureScript might be better choices for you.

(I've been writing a bunch of CoffeeScript lately and loving it. I'm very anxious to try my hand at ClojureScript also. GWT has always seemed very unwieldy to me.)

This is not a "Clojure thing."

The advice to noobs not to go hog wild with macros is common among experienced Lisp programmers in all dialects. Paul Graham himself refers to it in _On Lisp_: Chapter 8 is titled "When to Use Macros" and the first section is "When Nothing Else Will Do."

"my/our continuing work on Clojure is an ongoing gift"

and what a gift it is! Thank you Rich (and the rest of the Clojure community) for this wonderful language.

Personally, it makes my work more enjoyable when I use it and I look around and see folks all over having fun with it. Plus it's creating jobs and competitive advantage.

Not everything is awesome, but from where I sit, Clojure sure is.