iamaaditya

6 ms·

1.
by iamaaditya 9mo ago
One usage case that I saw myself is when a vehicle is parked such that it will require the other vehicle to go slighty over the curb, in this case the curb is flat so I assuming the parked driver thought it was okay. Every other human drive
2.
by iamaaditya 4y ago
In machine learning (especially deep learning or neural networks), the 'training' is done by using Stochastic Gradient Descent. These gradients are computed using Backpropagation. Backpropagation requires you to do a backward pass
3.
by iamaaditya 6y ago
Love the UI and the simple design. All the best with the product!
4.
by iamaaditya 6y ago
I believe there is some form of memory bias or bias in reporting by the media. If you look at the record books, you will find that most records are held by Sherpas. For example, more than 70 records for Mount Everest is held my Sherpas (or
5.
by iamaaditya 6y ago
Graham's Number, as in 3 ↑↑ 3, is named after Ron Graham. [1] https://plus.maths.org/content/too-big-write-not-too-big-gra... [2] https://en.wikipedia.org/wiki/Graham%27s_number?oldformat=tr..
6.
Features of a CNN overlap, reducing this leads to better generalization(arxiv.org)
2 points by iamaaditya 8y ago|0 comments
7.
by iamaaditya 8y ago
[The Matrix Calculus You Need For Deep Learning]( https://arxiv.org/pdf/1802.01528.pdf )
8.
by iamaaditya 8y ago
One simple way to minimize impact of these attacks is our work called Pixel Deflection (CVPR 2018 Spotlight). Here is a short (4 min) video introduction to the idea https://youtu.be/VgjOXJ9QKWo
9.
'05 Annual Performance Review: Albert Einstein(norvig.com)
42 points by iamaaditya 9y ago|6 comments
10.
by iamaaditya 9y ago
Paper: https://arxiv.org/abs/1801.08926 Code: https://github.com/iamaaditya/pixel-deflection
11.
Pixel Deflection – a simple transformation based defense(iamaaditya.github.io)
1 points by iamaaditya 9y ago|1 comments
12.
Neural Paraphrase Generation (CODE+DATA)(github.com)
1 points by iamaaditya 9y ago|0 comments
13.
by iamaaditya 10y ago
Technically it was heat but mostly it was due to (i) Economics, there was less demand for faster clock speed. Otherwise more research could have gone towards solving heat problem. (ii) Each cycle of CPU was more efficient with ability to ex
14.
Lecture Notes on Randomized Linear Algebra (2013)(arxiv.org)
163 points by iamaaditya 10y ago|18 comments
15.
by iamaaditya 10y ago
Recently, I have seen O1 being given pretty generously. Couple of people who I know who completed PhDs from mediocre University and less than 10 papers in total have all gotten their O1s. Yes they had to get reference letters and complete a
16.
by iamaaditya 10y ago
If it would be sufficient to have high PSNR only in the 'regions of interest', then you could use something like this [1], which uses a CNN model to predict a map, and a multiple JPEG encoding to achieve variable quality compressi
17.
by iamaaditya 10y ago
Here is my work, where I do use semantic information to achieve compression (rather improve JPEG). This is not an end to end compression like Google's work, but just incorporating semantic knowledge into compression. I am still trying
18.
by iamaaditya 10y ago
This library does not bundle Intel MKL. It is a prerequisite.
19.
by iamaaditya 10y ago
https://support.microsoft.com/en-us/kb/3053711 (from the link) Resolution To resolve the issue, do not create a user account contains the string "user" on the computer.
20.
by iamaaditya 10y ago
This reminds me of ensemble. Most machine learning techniques benefit in accuracy and precision by ensembling different methods. More different the methods are (variance) better improvement you get.
21.
by iamaaditya 10y ago
hi syllogism Right now everyone doing this is highly focussed on the competition and trying to beat the numbers. For that purpose certainly they would want to stick to predicting Top K answers. For e.g see this table Model Q+I [1] Q+I
22.
by iamaaditya 10y ago
1. Model Q+I [1] Q+I+C [1] ATT 1000 ATT Full ACC. 0.2678 0.2939 0.4838 0.4651 Where ATT Full represents using all the words in the vocabulary, as you can see it performs worse than "Most frequent 1000 answers&quot
23.
by iamaaditya 10y ago
So, I experimented with Skip-thought vectors (as question), to embody the the semantic knowledge about the question but that performed poorly, I mean very poorly. Tbh, I was surprised with that. Currently the way system works, is highly bia
24.
by iamaaditya 10y ago
Even though this is "Question Answering", it is trained as a classification model. Thus the model will try to come up with one of the top "1000" answers it has seen during the training. This certainly limits the possibil
25.
by iamaaditya 10y ago
Hi Harperlee I ran the question you asked -- "Are they playing tennis?" and following is the result 99.93 % yes 00.05 % no 000.0 % right 000.0 % left 000.0 % black and white