nharada

38 ms·

481.
by nharada 10y ago
It's no longer a common everyday saying because it's become synonymous with the sinister sexist undertones that it implies. It's like asking someone to "Netflix and Chill" today vs 3 years ago. At some point it may
482.
by nharada 10y ago
Any word on battery life?
483.
by nharada 10y ago
I don't understand this. You say you want more battery in exchange for a larger device. Isn't that exactly what this case is providing? You call it "monstrous" but it's precisely what you're asking for -- more
484.
by nharada 10y ago
don't forget https://www.ottoradio.com/
485.
by nharada 10y ago
Applying convolutional nets to one dimensional time series data is a fairly common problem in medicine, and the techniques here can certainly transfer to other medical domains. Besides the plethora of opportunities in one dimensional signal
486.
by nharada 10y ago
referrals and a PhD (usually)
487.
by nharada 10y ago
Nice, I love the interface to Keras. I wrote a similar application for Caffe but my interface isn't nearly as clean.
488.
by nharada 10y ago
Paper: http://www.mit.edu/~ilkery/papers/phys_nips.pdf
489.
by nharada 10y ago
and how do you work fulltime for a company and not bother to figure out what they do?
490.
by nharada 10y ago
The middle class STARTS at 150K?
491.
by nharada 10y ago
I can confirm they do ask homework questions that they could potentially steal work from. I interviewed for an ML position and was also given a fairly specific homework problem, and was not followed up on nor given information on how I fail
492.
by nharada 10y ago
It's crazy to me that Pebble manages to crowdfund most of its operations like this. I assume that since the delivery dates are so close they've almost finished development on the products? So really this more of a pre-order than a
493.
by nharada 10y ago
What architecture do you need to get 100% accuracy? Don't forget to use a validation set for the model and hyperparameter selection though!
494.
by nharada 10y ago
Couldn't you just downsample the wave file?
495.
by nharada 10y ago
Can someone familiar with the space speak to how much of this is advertising and how much is genuine progress/innovation?
496.
by nharada 10y ago
I'd also be interested in this too if possible, email is {hn username}1@gmail
497.
by nharada 10y ago
I'd argue that it'll soon be the other way around. Many of the improvements in Pascal were driven by scientific computing. It's easy to justify adding new features (like high speed 16 bit FP) when Facebook or Google is willin
498.
by nharada 11y ago
My understanding is that Watson basically acts as a cloud machine learning platform and people can build on it as they see fit. In that sense, Watson is basically the "brand" of IBM's machine learning offerings, no matter how
499.
by nharada 11y ago
Why do you think that other established players haven't entered this space (i.e. Boeing, etc)?
500.
by nharada 11y ago
Have you considered gaining weight?
501.
by nharada 11y ago
I agree, this is an insanely hard problem to solve. There are dozens of companies with tons of capital and massive R&D budgets that haven't already done this, presumably because they either think it's too hard, the regulatory
502.
by nharada 11y ago
I can't find a source for this, where did they say that?
503.
Hackathon Be Gone(brianchang.info)
268 points by nharada 11y ago|136 comments
504.
by nharada 11y ago
Anyone who has read this paper: how is their learning algorithm any different than a linear SVM?
505.
by nharada 11y ago
You could also argue that the money being taken by the state is being used for programs like public education and infrastructure instead of being stashed offshore by multinationals where nobody can use it.
506.
by nharada 11y ago
I was under the impression that specialized ASIC based miners have made even GPU mining relatively pointless.
507.
by nharada 11y ago
How do you mean "much better compression"? Won't replacing 32bit multiplies by bitwise operations save 32x the memory[1]? Han et al. show not only 35-49x improvement, but on much more difficult benchmarks (MNIST vs Alexnet&#x
508.
by nharada 11y ago
I too am unclear on whether or not the location tracking was related to the usage tracking. Usage tracking is basically the entire point of Amazon Underground.
509.
by nharada 11y ago
Relevant section of the abstract: > Adjusting for potential confounders, use of antidepressants during the second and/or third trimester was associated with the risk of ASD (31 exposed infants; adjusted hazard ratio, 1.87; 95% CI, 1
510.
Boston street light field guide(angelfire.com)
1 points by nharada 11y ago|0 comments

More ›