HNHacker News
TopNewBestAskShowJobs

neurangotan

10 karma · joined February 29, 2016

submissionscomments
neurangotan··on Ivermectin is effective for Covid-19: real-time meta analysis of 45 studies
>> Later a lot of evidence has been gathered that hydroxycloroquine is actually harmful

Where is this evidence ? Are you referring to this https://www.thelancet.com/journals/lancet/article/PIIS0140-6... ?

neurangotan··on Dynamic Memory Networks for Visual and Textual Question Answering
Thanks. That was the info I was looking for. Still a lot to digest but it is going to be really helpful.
neurangotan··on Dynamic Memory Networks for Visual and Textual Question Answering
Guys thanks both for the replies. I still do not understand the inputs and outputs of the attention mechanism neural network.

The paper says

eij = a(si−1, hj) is an alignment model which scores how well the inputs around position j and the output at position i match. The score is based on the RNN hidden state si−1 (just before emitting yi) and the j-th annotation hj of the input sentence.

We parametrize the alignment model a as a feedforward neural network which is jointly trained with all the other components of the proposed system.

So a is a feed fw nn. What is the input to this nn ? Do we input each hj corresponding to each word of the input sentence separately and get one number for each word ? If the input sentence has 200 words do I run this 200 times ? One for each input word ?

neurangotan··on Dynamic Memory Networks for Visual and Textual Question Answering
Hi thanks for a great paper. I have read it as well as the nvidia articles a lot of times but I am failing to grasp some important details. If you could shed some light that would be great.

The attention mechanism is a feed forward single layer neural network. The problem is that the input sequence is variable length. How are the outputs of the bi-rnn fed to the attention mechanism. What happens if I have a 80 word sentence as input and what happens if I have a 10 word sentence as input ?

neurangotan··on Greek startups to look out for in 2016
Not knowing what a linked list is, is like a math major not knowing what an integral is. It is a super basic data structure and an integral data structure to a whole family of functional languages. A computer scientist right out of university should at least have heard of the basic data structures used in computer programs.

Even if you don't know how to implement them you at least need to know how they work and their performance characteristics. I would not expect them to be able to implement one in C but at least they should what the fuck it is.

neurangotan··on Greek startups to look out for in 2016
Cyprus is pretty good for creating a company. They have low taxes and a nice industry of accounting firms revolving around this which makes things pretty easy. You can create a company without setting foot in Cyprus. Doing the paperwork for a company in Greece is a nightmare. Plus tax is 28% and constantly changes always upwards.

You are also obliged to pay for state health insurance in order to own a company which also changes constantly and upwards all the time. I you fail to pay your own health insurance you can go to jail. You cannot stop the health insurance. You have to close the company. If you owe health insurance payments and cannot pay them you cannot close the company and the health insurance debt keeps mounting up.

neurangotan··on Greek startups to look out for in 2016
Junior dev salaries with absolutely zero experience start at 800 net x 14. That translates to a total cost x 1.7 for the employer which is 1360. I am talking about devs that have never heard of a linked list. Most good people go abroad.