Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
zk00006
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
9 ms
·
31.
▲
by
zk00006
12y ago
There is a good reason to be negative a priory in real live. Consider you start a thorough discussion on any problem that you encounter during the day. You wouldn't be doing anything else and die of hunger. In my opinion, people must b
32.
▲
by
zk00006
12y ago
Given the fact that people get points for their answers, there is already bias towards positive and encouraging comments in hacker news. Notice how the comments start positive and later slowly add the information just to be "safe"
33.
▲
by
zk00006
12y ago
Is it possible to YAPF it in Sublime?
34.
▲
by
zk00006
12y ago
There are two key problems I am aware of. First, some countries such as South Africa banns UAVs so it is in fact illegal to deploy such systems. This was the key problem for the "Rhino challenge" from the last year. Second, it is
35.
▲
by
zk00006
12y ago
So this means that his reason must be personal and important only to himself. That leaves me wondering, what personal reason can one have to do such a thing?
36.
▲
by
zk00006
12y ago
Experts are in fact skeptical against fully self-driving planes due to safety reasons.
37.
▲
by
zk00006
12y ago
Honest question. Is this act of terrorism? If not why?
38.
▲
by
zk00006
12y ago
If you ever work with classifier training, these results are not surprising. You can take all false positives generate by classifier, average them and you will come up with an image that resembles the object to be recognized.
39.
▲
by
zk00006
12y ago
The whole idea is to tear down algorithm, simulate unnatural image that triggers the required responses and pass it to classifier. There is no direct link to security risks here.
40.
▲
by
zk00006
12y ago
You can not be serious about all possible research trends. CNN is one of possibilities that you can spend time on. The faster you get the key ideas the faster you can move on to something else, or dig deeper as you propose.
41.
▲
by
zk00006
12y ago
Visualizing training of CNN in browser allows to anybody quickly get the ideas behind. A while ago, I was looking for things like "xyz applet" to get intuition about xyz. This is exactly what is going on here. Great respect to the
42.
▲
by
zk00006
12y ago
Wow. You probably mean the "Hello." manual.
43.
▲
by
zk00006
12y ago
Lucky you. But this is a problem that should be solved independent of warranty. The quality of the display is clearly not reaching customer expectations.
44.
▲
by
zk00006
12y ago
Absolutely. Also claiming that the screen was damaged by bad cleaning method actually means that all my 5 laptops I had in the past are somehow super-strong but only Macbook Pro Retina 2013 is normal.
45.
▲
by
zk00006
12y ago
Great effort. Same issue here. It is clearly a problem in production of Retina. Any other computer i have from apple does not have this issue and is still OK while being older. I honestly feel embarrassed to display my own work to clients o
46.
▲
by
zk00006
12y ago
This list is too personal and I personally don't find it beneficial. If you are looking for introduction to computer vision in general, get a good book such as "Computer Vision Models Learning and Inference" or start directly
47.
▲
by
zk00006
12y ago
Nice application! Based on what I have seen so far, I am pretty sure they are using a tracker based on correlation filters [1]. These are very fast and robust. Naive implementation does not provide scale, which is visible in the video. but
48.
▲
by
zk00006
12y ago
Yes, same issue on retina Macbook Pro 15, bought 2013. Apple refused to repair the screen. It is clearly that the scratches match the structure of the keyboard. Previous model does not have this issue and is used identically. I bought the l
49.
▲
by
zk00006
12y ago
Funny how the article injects names of movies to the page to get higher page rank at Google. At least this was the strategy long time ago :)
50.
▲
by
zk00006
12y ago
Interesting idea. I wonder whether you wrote this comment by your phone or your PC though ;)
51.
▲
by
zk00006
13y ago
Based on the posts, people think that seznam.cz is a startup and Google will buy it like in 3,2,1. This is complete nonsense. Seznam is far from a startup and I am pretty sure their goal is not to get "only" acquired. Its mapping
52.
▲
by
zk00006
13y ago
Seznam.cz is far from being a startup.
53.
▲
by
zk00006
13y ago
Souhlas.
54.
▲
by
zk00006
13y ago
I work myself, so don't actually have much experience with this rule. But people need negative feedback which often creates big motivation for them and enforces learning. Giving only positive examples does not help to become better per
55.
▲
by
zk00006
13y ago
Honest question: can somebody explain what problem Keybase addresses?
56.
▲
by
zk00006
13y ago
To get to point when you don't need more money is not that hard. But it certainly depends how much you want. In the end you have one hamburger for lunch anyway [Bill Gates]. Attention is trickier. It brings more power, but also more so
57.
▲
by
zk00006
13y ago
Can VR displays replace monitors for programming? It would be great to have virtually unlimited screen size. Are there some problems I do not realize at the moment?
58.
▲
by
zk00006
13y ago
I believe these speed-reading tools are missing the concept of meaning. While the idea to fix the eyes allows to deliver more data to the retina, the linear speed does not work well. It would be great to combine these methods with some text
59.
▲
by
zk00006
13y ago
From my experience, I get the general meaning also by focusing on first and last sentence of a paragraph and skipping everything else. Also, thinking for a while about individual words gives a lot of information about the whole content with
60.
▲
by
zk00006
13y ago
Is it just me, or other people are also not able to get the content of the text while testing these method? Reading is highly non-linear and it is necessary to spend more time on some parts of the sentence than the others. Streaming the wor
More ›