|
Thread Rules 1. This is not a "do my homework for me" thread. If you have specific questions, ask, but don't post an assignment or homework problem and expect an exact solution. 2. No recruiting for your cockamamie projects (you won't replace facebook with 3 dudes you found on the internet and $20) 3. If you can't articulate why a language is bad, don't start slinging shit about it. Just remember that nothing is worse than making CSS IE6 compatible. 4. Use [code] tags to format code blocks. |
Anyone know about the current state of neural network sentiment analysis (language processing)? I am making a stock prediction NN, and I was thinking about merging what I am currently building with a sentiment analysis network - but im not gonna waste my time if it won't have reliable accuracy.
Available research on NNets for trading the markets seems to be pretty limited, for obvious reasons I guess.
|
On January 07 2020 10:56 Burned Toast wrote: Is there an API or something that can locate a specific frame in a Youtube video by image recognition ?
I would love to develop a more automated process when i write the posts for the for Small Vod Thread, and the ability to locate the frames with the image of the BW lobby in a long youtube video would be an interesting time saver...
Do you have some example vods to look at? One thing that comes to mind is 1) start with youtube-dl to download it to a local video 2a) Use some shot transition detection to write out shot detections with timestamp and screenshot, which would let you easily browse through the game start. See something like https://cloud.google.com/video-intelligence/docs/shot-detection or https://pyscenedetect.readthedocs.io/en/latest/ 2b) Instead of that, actually search for something. I'm not sure if you are only talking about vods that have bw lobby countdown (where maybe you can search for that by trying to modify some cv based stopsign detection code https://github.com/mbasilyan/Stop-Sign-Detection), or not.
I'm not up to date at all with the real best way to solve this so take my answer with a grain of salt.
|
|
|
On January 13 2020 14:31 JesusNolan wrote: How the programming language was created? Which one?
But essentially, start with Ada Lovelace and work forward in time. Although you could even argue that the basis for programming is with Leibnitz, or further back, al-Khwarizmi. But they didn't actually program anything, hence why Ada Lovelace is, imho, the first programmer.
|
It's a longshot that I will be able to get help here for this, I have done a lot of research and come up short.
I have a neural network made with keras backed by tensorflow.
At one point I have a symbolic tensor (output of a layer) that has shape (None,500,8). (it's none because keras doesn't define the batch size).
I have another tensor of shape (None,1), where every value corresponds to an index to slice from.
I want to dynamically slice from the first symbolic tensor using the second symbolic tensor. For each slice, I want to take 20 elements. So for example if my batch size was 2, and my index tensor was ([[0],[7]], then I want to extract a single new tensor from indices: [0,0:20, and [1,7:27,.
I can do this using a loop to look at each sample of my batch independents and then concatenating them together into a new tensor. But, I cannot seem to find a way to do it vectorized. It's actually really shitty, and makes me want to start looking at backends that aren't tensorflow because this seems super limiting.
|
I have a really dumb question that I attribute to my lack of formal CS training.
I'm creating some model validations for the system and I'm getting some weird conversions on floating point numbers...
Namely, when I pass 1.123 as an attribute it shows it as 0.112e1 and I have absolutely no idea why. Maybe Rails wants to be too smart and converts it based on actual db column settings?
# Column definition in the db (Postgres) bid numeric(8,2) DEFAULT 0.0 NOT NULL
# Validations for this column in the code validates :bid, presence: true, format: { with: /\A\d+(?:\.\d{0,2})?\z/ }, numericality: { greater_than: -1_000_000, less_than: 1_000_000 }
Any ideas on what could be wrong? 1.123 shows as valid, even though it shouldn't be and when I inspect the model it shows as 0.112e1. Column is defined as decimal with precision 8 and scale 2.
|
On February 04 2020 13:07 Manit0u wrote:I have a really dumb question that I attribute to my lack of formal CS training. I'm creating some model validations for the system and I'm getting some weird conversions on floating point numbers... Namely, when I pass 1.123 as an attribute it shows it as 0.112e1 and I have absolutely no idea why. Maybe Rails wants to be too smart and converts it based on actual db column settings? # Column definition in the db (Postgres) bid numeric(8,2) DEFAULT 0.0 NOT NULL
# Validations for this column in the code validates :bid, presence: true, format: { with: /\A\d+(?:\.\d{0,2})?\z/ }, numericality: { greater_than: -1_000_000, less_than: 1_000_000 }
Any ideas on what could be wrong? 1.123 shows as valid, even though it shouldn't be and when I inspect the model it shows as 0.112e1. Column is defined as decimal with precision 8 and scale 2.
This makes sense I think. Its a numeric data type with 8 precision and 2 scale. This basically means that you have 8 digits total in your number and there are 2 decimal digits. When you pass 1.123, it truncates the 3, as you are passing in more digits than the expected date type. It looks correct as 0.112e1(scientific notation) = 1.12 which states that there are 2 decimal spots.
I don't know the exact database you are using but here is the sql server docs on Numeric type: https://docs.microsoft.com/en-us/sql/t-sql/data-types/decimal-and-numeric-transact-sql?view=sql-server-ver15
|
It's what I thought but it just surprised me it does it for decimals but doesn't for other types (on the model, nothing to do with db yet). You can try and assign strings to integers or booleans and they don't get converted. That's why I wanted to add the validation so that the user knows they screwed up by adding too many decimal digits but it seems that I can't properly validate it now.
I'll see if I can work around it somehow and force it to not coerce the value during assignment.
|
Hey, here is a question for everyone:
Is it normal to need to look up how to use the same function in any given library over and over and over? Or am I just braindead?
|
Hyrule18771 Posts
|
|
Depends on the function, some just are not designed to be very memorable...
|
I'm pulling my hair out trying to figure this .NET MVC problem. In a view, I have the following EditorFor command:
@Html.EditorFor(model => model.Export, "EditorTemplates/Report/Export") For whatever reason, it does not find and use the specified template. I cannot think of any reason why. I've tried renaming the view and its subfolder. I've tried using this command in a different view. None of it works.
I know it doesn't work because I have a breakpoint in the specified view. Any ideas?
|
Hi enigmaticcam,
A little more information about the MVC version you are using and what controller you are in and the folder structure would have been useful. That being sad: - Have you tried giving the full path to the template (like "~/Views/Shared/EditorTemplates/Report/Export") perhaps with the ending ".cshtml" at the end? - Usually, when a view cannot be found, you will get a runtime error with the folders the view was expected to be in. Since you do not get an error, there might be another view with the apropriate name in a folder "higher" up the hierarchy. In this case your break point would not get hit because a different view gets rendered. - Have you tried cleaning the solution and recompiling it? This should get rid of some funny business (even though chaning views should work without having to recompile) - Last but no least: Did you compile and run the solution in debug mode? (a classic :D)
|
|
I'm looking at sparking some friends' interest in programming while we are all trapped indoors. Can anyone recommend any beginner online programming courses that use Python? Ideally free or very cheap and as high quality as possible. These are people with no real background for development at all, though they will have me to ask questions of.
I'm choosing Python because I'm familiar with it and I'm making the assumption that it's one of the easiest to learn.
|
On April 10 2020 01:30 WarSame wrote: I'm looking at sparking some friends' interest in programming while we are all trapped indoors. Can anyone recommend any beginner online programming courses that use Python? Ideally free or very cheap and as high quality as possible. These are people with no real background for development at all, though they will have me to ask questions of.
I'm choosing Python because I'm familiar with it and I'm making the assumption that it's one of the easiest to learn.
Obey The Testing Goat!
One of the best programming courses ever. It's Python + Django for web apps.
If they're willing to pay up then you have this for $30: https://learnpythonthehardway.org/python3/
|
Hyrule18771 Posts
|
I suppose I should add more criteria to my request:
This would need to be something entirely within the browser, with no installation. If you've seen w3schools Im thinking along those lines. It would also need ideally 0 background knowledge. Example from LPTHW:
If you do not know how to use PowerShell on Windows, Terminal on macOS or bash on Linux then you need to go learn that first. You should do the exercises in Appendix A first before continuing with these exercises. would just immediately filter everyone out.
These are people who have not programmed before. I am hoping they can get the basics down and understand how computers work a little better. I am not trying to make them into professionals.
I think I was also wrong in using the word course. Maybe more appropriate would have been tutorial or introduction. The Stanford course has homework, you need to apply to get in, etc. This would not work for this group of people.
|
|
|
|