dimatura

30 ms·

331.
by dimatura 7y ago
:) I forgot to say "our papers" -- credit also goes to the coauthor.
332.
by dimatura 7y ago
Looks like an interesting paper - I agree with your single-sentence summary, at least. I'll check it out.
333.
by dimatura 7y ago
Oh yeah, can't argue with results. Similar to deep learning. I've used both deep learning and metaheuristics a fair amount -- I don't care too much about mathematical rigor ;). I just mean, it's the sort of thing that us
334.
by dimatura 7y ago
I'm old enough to have started doing machine learning before the deep learning revolution. When it happened, I predicted we'd see a resurgence in metaheuristics (of which ACO, used here, is an example), which just like neural netw
335.
by dimatura 8y ago
I do it as a way to deal with perfectionism. I write program0 (hey, this is computer science, we start at 0) and often never get around to program1, since program0 works, despite being a bunch of hacks. But I feel less bad about the hacks w
336.
by dimatura 8y ago
I have felt the same way in the past. There are some ways to do this, but none is great. Unfortunately using a text editor to put in the notebook text areas is not that straightforward because of security features in modern browsers. Since
337.
by dimatura 8y ago
Yes and no. I do computer vision and like photography as well. I use Jupyter notebooks extensively for computer vision, and they work pretty well for (semi) interactive manipulation of image data with code. But as a general purpose tool, it
338.
by dimatura 8y ago
Yes, it's basically the idea in Duff's device. Protothreads ( http://dunkels.com/adam/pt/ ) uses a pretty similar trick.
339.
by dimatura 8y ago
Super useful, in fact. A lot of the code I write follows a typical pattern of bottleneck code in C++ with python wrappers, or a mixture of pure python nodes (for the things that don't have to be super fast) and pure C++ nodes. This is
340.
by dimatura 8y ago
catkin_make and catkin-tools (where "catkin build" comes from) both use catkin. They're just different CLI tools, and for the most part compatible, and have been fairly stable for about four years now. How many build tools ha
341.
by dimatura 8y ago
ROS started development almost 10 years ago. A lot of "high quality" solutions that seem like a good idea now may not have been around that long ago, or were still in nascent stage. Many of the changes in ROS2 are actually about i
342.
by dimatura 8y ago
CMake is pretty established. You don't have to use CMake and if you don't want to, you don't have to use it - you can just compile your code however you like and manually put all your package files where ROS expects them to
343.
by dimatura 8y ago
You can use ROS with Python and Linux. In fact, that's a big chunk of actual ROS-based systems out there, with most of the rest being C++ and Linux. For simple robots, you can get away with a simple script. But say, you want to attach
344.
by dimatura 8y ago
What is the "pompous deep learning bullshit"? They don't really dwell a lot on DL itself, just describe a reasonable architecture for the task and how they train it. While it's possible to conceive the same concept reali
345.
by dimatura 8y ago
Sorry, but I think you're fundamentally misunderstanding the idea of the paper. Colorization is not the point - it's an auxiliary task, that lets the algorithm discover how to do a form of tracking. As the paper itself states, the
346.
by dimatura 8y ago
PyContracts ( https://andreacensi.github.io/contracts/ ) supports NumPy arrays.
347.
by dimatura 9y ago
The whole project is tongue-in-cheek, but it's real and it works. Disclaimer: I'm a contributor ;)
348.
by dimatura 9y ago
Agreed - I don't enough to say about this particular project, but the lasers in LiDAR sensors do have the potential to be powerful enough to harm eyesight, so I would advise caution. This is not a concern for most commercial sensors, a
349.
by dimatura 9y ago
Does it do filtering over time?
350.
by dimatura 9y ago
I absolutely love lisp. It's crazy how it's one of the oldest language around, yet it still has more features than many other newer mainstream languages. At the same time, one of the reasons I don't like it too much for daily
351.
by dimatura 9y ago
I've used moviepy quite a lot and even submitted a PR or two. It's not perfect, and docs aren't super up to date. For many tasks a good nonlinear editor is way easier. BUT, it also makes some tasks really easy that seem surpr
352.
by dimatura 9y ago
Nice, I'm a big user of various data stores for scientific work. How does this compare to LMDB ( http://www.lmdb.tech/doc/ ), which uses B-trees (and has a Python interface)?
353.
by dimatura 9y ago
I work in the field. Basically, yes. On average it is easier to fly than it is to drive, mostly because there's far fewer things to crash into. But on the other hand, almost any failure can become catastrophic - falling out of the sky
354.
by dimatura 9y ago
Ah, that's an interesting shortcut.
355.
by dimatura 9y ago
I write in a mix of python and C++ as well. As such, I've also gradually moved towards using conda. What's your dev workflow for the code that's not a dependency like? Specially if the code you're developing is C++. Do y
356.
by dimatura 9y ago
How is an analog computer "easier to program" than a digital computer? Making neural networks do what you want is hard enough with the help of tons of libraries, decent scripting languages, the ability to dump the weights into a f
357.
by dimatura 9y ago
I tried to do this once with Theano, and found that the latency of the roundtrip to GPU and back made it not worthwhile for a single image. Maybe a batch of images at once would make it worthwhile. And this isn't what theano is intende
358.
by dimatura 9y ago
I've been using tiling window managers for over 10 years, and for this time has been spent on ratpoison (after brief stints with wmii, awesome and dwm). Despite i3/xmonad/bspwm/etc/etc I keep going back to ratpoiso
359.
by dimatura 9y ago
Seems like out of the major libraries (TF/Caffe/Theano/pytorch), pytorch is the only one to have a core that is C (the TH*). It's not exactly a small library, though. One small library that is in C and has some state-of-
360.
by dimatura 9y ago
One thing that seems promising is built-in support for binary neural networks, which makes sense given its focus on embedded devices. No reason this couldn't have been implemented, in say, pytorch - but I'm guessing this library w

More ›