Hi, author here. I love reading academic papers and I love Math. But it's an unfortunate fact that academic papers make up the majority of the literature on ML — and that their notation, and writing style exclude a large population who would otherwise find them useful. That's all I'm trying to get at here.
They don't exclude people by using math any more than programmers exclude people by using subclasses or anonymous functions.
I too would love to see more entry level ML resources, but we are talking about academic papers here.
Imagine how stupid it would seem if some outsider came in to HN and started telling programmers to dumb down their code when it's open source because their advanced techniques are too hard for a novice to understand off the bat. Instead of optimizing for efficiency, elegance and readability for their peers - the people they are collaborating with to solve actual problems - they're told to cater to novices always.
The language in your post is textbook anti intellectualism isn't it? And strangely entitled. You would certainly not apply these criteria to code but since it's not your field they must cater to your lack of experience? You know better than they how to communicate ML research to other experts?
> Aside from the fact the techniques he mentioned are some of the simplest in machine learning (and are hardly those that would immediately come to mind when I think "machine learning")
The primary point of this article was the contrast between something as simple as K-Means, and the literature that describes it. It wasn't meant as a full intro to ML, but rather something along the lines of "give it a try, you might be surprised by what you can achieve".
> Even for "simple" algorithms such as K-Means, implementing them is often only the tip of the iceberg.
Yup. But getting more people to explore the tip of the iceberg is, in my opinion, a good thing. We don't discourage people from programming because they don't instantly understand the runtime complexity of hash tables and binary trees. We encourage them to use what's already built knowing that smart people will eventually explore the rest of the iceberg.
Thanks for responding. I fully agree with your comment -- as I said, I too think many people are sometimes put off by the apparent complexity of machine learning, and demystifying how it works is a great thing.
Unfortunately there's always a risk that a "hey, maybe this isn't so hard after all" might turns into a "wow, that was easy". While I think the former is great, the latter is dangerous because machine learning is often used to make decisions (sometimes crucial, for example when dealing with financial transactions), so I would argue more care should be taken than if we were talking about general purpose programming: if you trust an algorithm with making important business decisions, then you better have an intimate knowledge of how it works.
While I again agree with the underlying sentiment, I was just a bit disappointed that it seems to invite the reader to be satisfied of himself rather than motivate him to dig deeper. Nothing a future blog post can't solve though!
The point of the graphs are as an example more than anything — in any real world solution it's not trivial to plot your documents (as representing documents with thousands of words in 1d, 2d or 3d is not trivial).
But if _you could_ represent your documents in 2d, you could plot them like I do here. And, the X-axis would represent 1 feature of your documents, and the Y-axis another.
Tl;dr don't focus on the axes — focus on the idea of being able to compare the relative distances between documents.
Yup. We're working on making this clearer right now.
"Your users" is _your_ users. You add a Javascript snippet to your app which sends us up-to-date information about your users (as they move around your app) — we record that data and allow you (or a colleague) to download a CSV of the data at any time.
In addition to adding the JS include for your script, we need to call a series of EasyCSV.user functions to identify the user currently viewing that page and set the data we want to store about that user for that page. But all that data we'd need to pump out into the JS calls is already coming from somewhere... a DB I would assume. For me, that seems like a bit too much overhead for such a simple task. Since all the user data is in the DB already anyway, when we want a CSV of our users, we do DB export. Unless I am missing something, I don't see this being very useful myself. But if it helps you, then it might help someone else.
I'm often asked to generate CSVs of our user base for various non-technical colleagues. They might use these CSVs to do analyses in Excel or import users into a service (like Mailchimp).
This is an attempt at taking me out of that process, and automating it.
I think there's a big difference between how [my generation and] I (19 y.o.) use Facebook, and how older generations do.
For me, it is 80% keeping up with my current friends and 20% older friends. I think these ratios are perhaps flipped after a certain point, perhaps the people for whom their FB graph wasn't created organically from the age of 14/15 but rather, had to go and create it posthumously.
Yes, there are people I'm friends with who I'm not friends with anymore. But, if you use Groups + Lists (which all my non-techy friends use too), this isn't really an issue.
Anyway, I just see a lot of commentary on things FB is doing wrong by people > 25/30 y.o. But, for people below that age bracket (who tend not to be commenting in the media) it has (from my observations) replaced texting/phone/skype/letters/etc.
I think there's a real divide between the people who use Facebook, for whom it is (often) a serious part of their social life and, those who comment on FB in the media and invest in FB on the public market.
Being 19 Facebook is a critical part of finding out what's going on, organising events and, keeping in touch with everyone (whether I see them every day or once a year). I'm not unique in this, it's true for all my friends and pretty much anyone I meet, of my age, around the (western?) world.
Obviously, the market thinks FB is over-valued; I'm not commenting on whether it is or isn't. I'm just interested at what the market might value FB at if the people buying its stock used it as much and in the same way as younger generations. (As far as I see it, it's like a single private company owning the whole "phone system" for young people. What would that be valued at?)
For me, the valuation of FB is closer to a TV channel with an audience of c. 500m people for 30mins every day of the week. And one which shows ads to people based on their own interests. What's that worth? I'm not sure.
The comparison with a phone network isn't quite right because I can't see FB ever being able to charge for service.
But I agree with your point. I think Linkedin is far more likely to be wiped out than Facebook, but the market thinks different.
I like the TV analogy, it's perhaps more apt. Surely the point of any TV network is to show ads based on their viewers interests? Just FB is able to be much more targeted (at the same time they also have to compete with the huge over-supply of online display-ads, are my eyeballs worth more on facebook.com vs. joes-blog.com? I don't think so, which is the major issue with their current monetization strategy (EDIT: and they haven't shown that their targeting is significantly better than Google Adwords as of yet)).
I am confident though that they will find a way to make FB a seriously viable business. The list of things I use FB for is just so long, I don't see how they couldn't find at least one v. profitable workflow to tap into (organising a night out, checking-in on the night out, posting photos the next day, etc.).
There are things like adblock which allow you to fully experience facebook sans ads.
Yes, there is tivo etc, but you have to pay extra to get out of the ad experience (and it doesnt always work). TV ads are much more intrusive than FB ads, and I'd imagine TV-style facebook ads (like some websites that force you to sit through an ad) most likely will kill the platform.
Are you sure about that re LinkedIn? Given its narrower customer base and usage model, LinkedIn is already employing a freemium model and charging for a higher service level. Their actual users do constitute a revenue stream in a way that Facebook's never could.
"Facebook is a critical part of finding out what's going on"
" I'm just interested at what the market might value FB at if the people buying its stock used it as much and in the same way as younger generations"
A good company doesn't necessarily make a good stock. There's a real possibility that FB will become the next MySpace, especially if another firm figures out how to build a mobile social network and doesn't sell out to FB or Google.
Users are using a free service that they enjoy and want to continue to be free. Those who invest in Facebook want to make money because they're risking money. You have faith that they'll figure out their adwords, but are you putting your money where your mouth is - are you (or your friends) risking thousands or millions of dollars on your faith?
This really isn't anything new or unique to Facebook though. Customer satisfaction doesn't automatically make something a good investment.
The question isn't whether it's useful to a lot of people or not but rather how can FB make money from all these people? The infrastructure to support almost a billion users costs a lot of money. If FB was a subscription service with such a large user base, I think it would be valued quite a bit differently than it is as a free service.
I don't know. Obviously this matters in determining how much FB is worth. It probably isn't much right now but, I see the potential for huge future earnings, they haven't found their adwords yet, I'm confident they will.
