Viewing profile — snikolov

a program can still generate the score that a musician plays sincerely

People who do hard work tend to bite off more than they can chew, especially at MIT, and there is obviously a limit to how much you can handle. What MIT does right when it comes to…

It is online learning, but as far as I understand, there is another side to it. Right now, if you get a new data stream, you have to play around to build the right models for it (m…

It sort of comes down to the question of what's really being learned here? That's a great question. We are learning to recognize trends and non-trends based on previous examples. S…

Thanks! I'm afraid my background, like yours, is more in methods that are not specialized for time series, and so I couldn't credibly give any comprehensive references. My understa…

You bring up a great point. To classify something as a trend or not a trend, we have to use this artificial black box to supply ourselves with examples of what's a trend and what i…

You can get the trending topics through the Twitter API https://dev.twitter.com/docs/api/1/get/trends/%3Awoeid But I think what you are asking is how such a method would come up wi…

Not in general. I was part of the VI-A program ( http://vi-a.mit.edu/ ) at MIT, which allows you to do your thesis at a company.

I agree with you --- this seems like a perfectly good way to tackle the problem of trend prediction directly. What we had in mind was something that would be more generally applica…

Thanks for the interest! We collected tweets by sampling a small percentage of all tweets in a time window, to emulate what one might get from the streaming API (I did this as part…

My problem is often in the other extreme. I tell myself "I don't want to bother so and so, I still haven't tried x." But there's a point at which you should ask for help if you're …

Dynamic Time Warping is a pretty neat application of dynamic programming that takes two time series and tries to squish one time series to look like the other, with some constraint…

Indeed, projection onto the principal subspace is a kind of regularization, which makes for better generalization. This kind of makes sense intuitively. These slides go into more d…

For what it's worth, I loved your post too. It reminded me of chats with my advisor. I'd walk in thinking "who the hell am I to push the boundaries of human knowledge?" and walk ou…

A nonparametric model for anomaly detection, classification, and prediction in timeseries. It observes the trajectory of some measurement over time and compares it to tons of histo…

I would caution against this approach. First, he is aiming to pass all the course requirements for a CS degree from MIT in only 1 year (including the non-CS requirements). His defi…

What struck me most is that this course helped form lasting communities who are now studying other material together. That is just wonderful. A community you can discuss the materi…

I wonder why they couldn't (or didn't choose to) partner with individual professors on a per-course basis instead of entire institutions whose professors might not all be intereste…

According to a new analysis, if you want to run a tech company, you shouldn't drop out of college. More like "if you want to run a tech company, but you're not already running one …

I agree that that the boys and videogames thing is a bit of a misconception. But rather than attacking that misconception, we should be asking how poeple might come to have such mi…

Look at this way: for a 20000 category classification problem, guessing randomly will give you 0.005% accuracy. Compared to chance, this is pretty good.

Clustering algorithms operate on features, which typically have to be designed by hand. The appeal of deep learning is that it discovers good features automatically.

Would you care to elaborate? Surely most of the lines of code are not directly implementing theoretical things, but I would say the meat of the work is without a doubt theoretical.…