HNHacker News
TopNewBestAskShowJobs

romaniv

2,750 karma · joined February 2, 2011

submissionscomments
romaniv··on Some Starting Points for Deep Learning and RNNs
There is more to AI than complicated neural networks. There are simple well-studied algorithms that deliver predictable results and work very well if you apply them to the right problem. Look into K-means clustering, linear regression and decision trees to get the taste.

Good introduction to the field: http://ocw.mit.edu/courses/electrical-engineering-and-comput...

romaniv··on Jobs Involving Routine Tasks Aren't Growing
It's all fine and well, as long as all "non-routine" jobs don't require a Master's degree and 10 years of hands-on experience.
romaniv··on Jobs Involving Routine Tasks Aren't Growing
> web/enterprise bubble

That's 99% of all programming.

romaniv··on Go Game Guru – Learn all about the board game Go
While we're here, does anyone know a good Go app for Android? A lot of them look like spyware, judging by permissions.
romaniv··on Go Game Guru – Learn all about the board game Go
>For instance the computer may ignore a capture to play elsewhere and it can be hard to work out why.

I think the biggest insight I had about Go when I was learning to play it is that the game commences on the entire board at once. At the beginning I was over-concentrating on a single "active" region. In reality, abandoning some "battle" and adding stones to an "unrelated" region is a perfectly sensible strategy as long as you think it will give you territory in the long run.

romaniv··on Deep or Shallow, NLP is breaking out
> ut, really, it doesn't matter. Google has the airwaves (so to speak). No matter what happens anywhere else, in academia or business, their stuff is going to be publicised the most and that's what we all have to deal with.

One thing that bothers me about ANN hype is that those models are very resource and data hungry. Which suits Google perfectly, but will keep that kind of AI out of people's hands/computers for decades to come.

romaniv··on AlphaGo Is Not the Solution to AI
Are you implying that AI research in the past decades didn't have any "real" results? This is a popular meta-narrative these days, but it's mostly a product of marketing that tries to magnify the significance of current work at the expense of everyone who worked on similar things in the past.
romaniv··on Sources of ICQ desktop client by mail.ru
This.

I miss ICQ. Well, I still use it occasionally, since it's part of my MirandaIM setup. But I miss it whenever I use Skype for IMs. For starters, I can aggregate ICQ, Jabber and a bunch of other services inside one lightweight open-source client. Also, it had an entirely different social vibe, which is missing from Skype. It was okay to chat with near-strangers in ICQ. It was okay to add people from some random forum to your contact list. It was okay to say "hi" to people when they went online (as opposed to always-online on 10 different devices Skype mentality).

romaniv··on On the growing, intentional uselessness of Google search results
A newcomer can index, store, search and display data differently, but first they still have to get that data. Which means dealing with JavaScript, paywalls and other fun stuff like that.

Besides, regardless of what you do, you would need to have tons of storage, bandwidth, CPU power and a high-availability infrastructure.

romaniv··on Google launches voice typing in Google Docs
Can't wait till Google starts data-mining background noises to "improve" your experience. Yes, they can do it with other of their services, but this one is the most likely to provide the most personal information. Also, it will capture much more information, since you will be running it over longer periods of time.
romaniv··on On the growing, intentional uselessness of Google search results
Alas, it’s 2016 and there’s still no serious competitor to Google.

And there will not be any time soon. Writing an efficient crawler for what we call the "modern" web is not something a small or even median-size company can pull off. Google enjoys a tremendous competitive advantage: people specifically optimize webpages for what it can and cannot do. So any newcomer to the field will have to replicate tons of technologies Google had years to perfect (in addition to solving problems like storage, search logic and bandwidth management).

romaniv··on Why I love Node.js
Most developers don't control their tool-chains. Plus, some tools act like black holes for software. They bend all the infrastructure around themselves and there is no coming back once you cross the adoption event horizon.
romaniv··on Getting Deep Speech to Work in Mandarin
I keep reading about these algorithms that are "better than humans". Perfect image recognition, perfect speech recognition, parsing plain text-queries and answering questions, etc, etc. So where are the practical implementations?

All the speech recognition engines I've interacted with so far were awful. Not just bad, awful.

>Collecting such data sets could be very difficult and prohibitively expensive.

Uh, movie subtitles?

romaniv··on SoundCloud could be forced to close after $44m losses
I would be happy to pay something to Soundcloud as long as it's not per-song. It's a good service. Really easy to use as a musician and great for finding new music as a consumer.

Hm, maybe they should be selling some "audiophile" subscriptions with ability to tag and hierarchically categorize songs you liked? I would buy that if the price is not astronomical.

But I also think it's really important that the basic usage stays free for people who randomly wonder in from other websites.

romaniv··on Awesome machine learning
Not sure it's a bad thing. Yahoo worked when the internet was smaller. For fields with reasonable amount of information the model is better than just searching for stuff.
romaniv··on Go, Marvin Minsky, and the Chasm That AI Hasn’t yet Crossed
Why 'seemingly'? Why is that not 'actual' groundbreaking progress?

Because mapping problems to pre-existing algorithms is bread and butter of computer science and software engineering. To be groundbreaking a work needs to change our understanding of the underlying issues. In a lot of popularized cases that does not happen.

romaniv··on Go, Marvin Minsky, and the Chasm That AI Hasn’t yet Crossed
Framing questions about AI in terms of "mastering" some field is somewhat disingenuous in the first place.
romaniv··on Go, Marvin Minsky, and the Chasm That AI Hasn’t yet Crossed
There is another way of saying the same thing. A lot of seemingly groundbreaking progress in AI usually happens when people (people, not machines) discover a clever way of mapping a new unsolved problem to another - well-solved - problem. That's a valid and insightful observation that you shouldn't dismiss with such an ease.
romaniv··on Pikazo
You're not making an argument that Pikazo generates real art. You're making an argument that humans don't make real art. This notion has absolutely no interesting implications for AI. It's like saying "language is relative, so any output of Google Translate is just as valid as what you think is valid".
romaniv··on Pikazo
Question for someone who has this app. Can this convert a photo to anime-like or cartoon-like rendering a-la A Scanner Darkly?
romaniv··on Pikazo
If art for you can be replaced by this kind of mimicry, it begs the question of whether you really need or care about art in the first place.
romaniv··on Pikazo
If a computer generates 100000000 unfunny jokes and one that seems funny, does that count?
romaniv··on Storing Passwords in a Highly Parallelized World
I do not think ACM article is very relevant here, because what is speaks about is, essentially, the fact that what we traditionally thought of "disks" is now getting as fast as RAM, because persistence is achieved via new technologies.

A machine with a measly 256GB of memory can do around half a million Argon2 hashes simultaneously

Not necessarily. There is always the limitation of memory bandwidth. Although, I do not see saturating that bandwidth as a design goal of Argon2.

romaniv··on When will librarians start to throw offline literature away?
A library without physical books is called a server.
romaniv··on Open Letter to Mozilla: Bring Back Persona
> Anyone with access to your email account can simply reset any password on any site. The right solution is to make your email account very, very secure.

No, the right solution is to stop using email as sole identification for password resets. Yes, there are other solutions you can implement right now without waiting for some big company to save you.

The most obvious one it to create a second factor of authentication just for account resets. It could be via an SMS OR simply by asking user to print out/write down a special randomly generated "reset" number.

"But SMS costs money!" No, for most providers you can send an email to a special address reserved for the phone number. It will get translated to SMS automatically.

"But users will forget/loose their reset number!" Maybe, maybe not. It's a cultural thing. You don't expect them to loose access to their email, but that happens all the time.

romaniv··on BetterExplained: Math Lessons That Explain Concepts
After reading several articles I came to the conclusion that every Wikipedia article on math concepts which are covered on betterexplained should start with the link to betterexplained. I know it's not going to happen, but it would benefit hundreds of thousands of people.
romaniv··on What It's Like to Be Noam Chomsky's Assistant
Out of curiosity, where did you hear about that research on false memories?

Also, how do you know you can rely on your memories of yesterday or memories of things you've learned in school?

romaniv··on A visualisation of all of the money in the world
You're comparing a currency + transaction network with a commodity. IMO, as long as people treat Bitcoin as a commodity it will not achieve its true purpose.
romaniv··on The first person to hack the iPhone is building a self-driving car
We have millennia of experience in regards to estimating how people will react to various shock situations and what constitutes those situations. It's intuitive.

The reactions of NN to unusual stimuli are likely to be counter-intuitive at best and unpredictable at worst.

romaniv··on Death by a thousand likes: Facebook and Twitter are killing the open web
Paywalls do not work, but Kickstarter and Patreon do work.

Also, hosting a website for several thousand people (as opposed to hundreds of thousands) it so cheap this days that the cost becomes negligible. It maintenance becomes the most "expensive" - or annoying if you're doing it yourself - part of it all.

← PreviousPage 14 of 30Next →