hackpert

6 ms·

61.
by hackpert 9y ago
This is amazing. I don't know if a lot of people here realize this, but it is really hard to pull off this level of integration of different computer vision components (believe me, I've tried). Microsoft has really outdone themsel
62.
by hackpert 9y ago
In addition to the brilliant accessibility features, there's a limited set of devices running iOS, all with known camera specs and picture quality, which can definitely help in maintaining some consistency in results by reducing variat
63.
by hackpert 9y ago
More than anything, I am excited about the possibilities this opens up. A lot of fields that DARPA touches with its funding-filled hands turn into a technology that defines the future, from the internet to self-driving cars to GPS. Exciting
64.
by hackpert 9y ago
This is great. Using HTML5 games in a headless browser makes a lot of sense because the need for VNC is circumvented. However, I think that while OpenAI's implementation is certainly not the best, having access just the information on
65.
by hackpert 9y ago
I'm interested in seeing how computationally efficient this method turns out to be and how well it generalizes to other audio data and perhaps other signals as well. Going on a hunch by the model, I think there are some more efficient
66.
by hackpert 9y ago
I'm curious why you think Abbeel is underrated. He is one of the few ML researchers to not succumb to the pull of the industry (at least completely) and his work is highly respected in the community, especially continuous control.
67.
by hackpert 9y ago
Sure, it is actually quite easy to design systems that allow only the state or the government (or any given trusted organizayion for that matter) to decrypt secure communications. However, the problem is, can we trust the governments to no
68.
by hackpert 9y ago
This transition from end-to-end differentiable 'black box' systems to multiple networks dedicated to certain tasks working in conjunction is interesting and very probably the idea that will keep this field going. We might not full
69.
by hackpert 9y ago
Couldn't agree more. A (somewhat large) company I worked with started working on this tech about three years ago, and still haven't gotten around to even coming close to launching it.
70.
by hackpert 9y ago
This is brilliant. I might actually consider switching to Windows now. Both Arch and macOS have been getting on my nerves lately.
71.
by hackpert 9y ago
For (1), you might want to look into http://dontprint.net . Of course it doesn't solve annotation issues but for reading it works well enough.
72.
by hackpert 9y ago
Yeah. I mostly end up using p2.xlarge spot instances, and they have worked fine for my jobs so far. Be sure to checkpoint often and you'll be okay with spot instances.
73.
by hackpert 10y ago
I'm interested and happen to have some prior (largely hobbyist research) experience in the field
74.
by hackpert 10y ago
Nice work though
75.
by hackpert 10y ago
incoming electron port in 5..4..
76.
by hackpert 10y ago
I'm curious, how hard do you think it is to integrate new tech with the systems currently being used in radiology? Are the file formats and standards largely proprietary/closed?
77.
by hackpert 10y ago
A friend and I tried to make a similar detector for removing squeaking sounds of whiteboard markers from videos. Although solutions like this do sound like a waste of time to some, I think they can go a long way in removing those tiny annoy
78.
by hackpert 10y ago
I think you're right about Penrose tiles being easy. One could, however, potentially use this with polygonal tiles to find their Heesch numbers and maybe even make some advancements in solving some of the many unsolved problems associa