vivekn

The problem with doing that is you keep reinventing the wheel. Use deep learning if that fits the problem, but keep learning about other things like how EM works, variational inference, graphical models etc on the side. One day you might find a problem where deep learning doesn't work as well as some of the other techniques. Sure there are data science jobs that can be done without much knowledge, but people tend to stop when they see math and are just happy to use some API. This IMO is a wrong approach.

Well MOOCs may not appeal to everyone but they are extremely useful to anyone who wants to learn about certain advanced topics on their own. It will not disrupt college education, since its not the courses or content that matters but the credentials.

What if you don't set the "Server" header?

Hmm, they should have handled negations, its not that hard to implement.

The model was trained on an English data set, but if you train the same model on some other data, it can handle other languages as well.

While 95% accuracy would be a really phenomenal achievement, an accuracy in the range of 85-90% is achievable using methods simpler than deep neural nets. I have done some work on sentiment analysis in the past. I used a Naive Bayes model with some enhancements like n-grams, negation handling and information filtering and was able to get more than 88% accuracy on a similar dataset based on movie reviews.

You can find more details here -http://arxiv.org/ftp/arxiv/papers/1305/1305.6143.pdf and the code over here - https://github.com/vivekn/sentiment/blob/master/info.py

Will be interesting to play around with that dataset.

Thwarted expectations and sarcasm are difficult to handle. You are down to the choice of optimizing for this particular case or the majority of text.

Nassim Taleb's "Antifragile", Fyodor Dostoyevsky's "The Brothers Karamazov", Robert Greene's "48 Laws of Power" were some of the good books I read this year.

I use wombat too, its easy on the eyes and a very aesthetically pleasing color scheme.

I was kind of impressed with dotcloud beta, its really simple to mix and match databases and deploy them. But I guess I can't afford $100/mo for hosting my kind of experimental apps. I was expecting a usage based pricing scheme similar to Amazon or ep.io. You guys really need to have some intermediate plans. Its a great product, by the way!

Thanks for your views.

As of now the 'details' are geolocation details. By clicking on a point on the map, the address and place info, if available would be added to the status update, that would be shared with the user's FB friends on his or her feed.

I am an EE student but with a lot of interest in CS . I personally am no big fan of C,despite the fact that most of the courses are in C . It is kind of cool when working on compilers or embedded systems but for anything else I would any day prefer something like Python or Haskell - the speed of development and the fun involved being the two main reasons.