I'm trying very hard to understand this code about a Deep Learning-Based NOMA system based in MATLAB. I am really new to MATLAB coding but I really need to understand this entire code as it will help in my school project and I am struggling.
I think as of right now I do not need to know how the mathematical formulas work, but instead, the focus is on what the code is doing and its flow.
This is part of the code in the trainData.m file that I am struggling with right now
Why are the pilot symbols calculated and then replaced right after?
Why is the idx_sc (20) selected to be replaced? What is its significance? Is it the only subcarrier selected for the training of the DL model? Why only that?
This portion of the code in the picture is labeled "generate training data for each class". From my understanding, it is generating OFDM packets for each label, simulating the transmission and reception, and then getting the features and labels for each of the 16 classes. Is that correct?
The code and all relevant function files can be found in the link below.
Please help me understand the code!!! Please! Much thanks!
https://www.mathworks.com/matlabcentral/fileexchange/75478-deep-learning-for-signal-detection-in-noma-systems
To get you started, In lines 91 the code initializes the entire variable as 0. Subsequent lines (92-96) are just replacing pieces of the variable based on the indexing inside the “(…)”
Related
I have been trying to use optimizer(SGD, Adagrad) from BigDL library on TransE with scala. My current implementation works with mini batch in sequential way. I followed this example to optimize the embeddings(as Tensors) without creating a layered model.My code is somewhat similar to this example. My current problem is, with some parameters my losses gets at a plateau point (the value of margin) no matter how many epochs I run. With this, my hit#10 in testing is not that good. Can someone give any idea why losses get at a plateaued point and if this generates bad testing results?
P.S. I have checked my loss calculation and it is fine. The only place I have control over my implementation is the optimizer.
Thanks in advance.
I have been researching on how to program image processing for counting objects and I found the following homepage about Matlab for cell counting
I am not familiar with Matlab, but found their ideas interesting, so reading this, I see that they are using a BlobAnalysis Object to find the centroid of the segmented cells (You can see the image :
Now, this seem very interesting but I wonder what this operation is doing exactly? (please don't give me the definition written in the docs-"it is finding the property in the blobs". How? is it segmenting it? separating it? ) As far as I can see either you separate the cells first (that is the whole point of counting- I am doing this in other program using watersheding) and then finding the centroids is just something added and not so important, OR somehow this blob analysis is doing some interesting segmentation itself that I would like to know.
Anyone familiar with this can give me some pointers or advice here?
I created a subsystem in Simulink with mask underneath. There are all sorts of control and calculation inside this subsystem. Now I have to duplicate this subsystem for one hundred thousand times because I need to connect one hundred thousands of this block in series.
What I have tried, I used the commands “add_block” and “add_line” where I can just type it in the Matlab command and the blocks and lines are added automatically.
What I wish to do now is,
I want to have 100 signals in a single subsystem, so instead of using one hundred thousand subsystem, I will only need one thousand of this subsystem, I understand that this can be done by vectorization.
I have a very limited knowledge on using vectorization feature in Matlab/Simulink. I appreciate if anyone of you could provide me a great reference on how to do this?
What I found here is something like this which I could not link it to my issue above: http://www.mathworks.co.uk/help/matlab/matlab_prog/vectorization.html
The other thing I found is by "using vectorization for most components. Most components are vectorized if they have a vectorized input signal or if one of their parameter is specified as a vector."
However, I could not find any further information/details, appreciate if anyone of you could give opinion on this? Thanks!
I am working on detecting road lanes using hough transform in simulink(matlab). I have designed the model but its not detecting the lanes. As i am very new to simulink so i have no idea where i am doing the mistake. I tried to upload the model but its not working. I have used the same model as given in matlab-simulink demo except i added a new block of rgb2intensity just after the image from file block and the image which i have given as an input is also uploaded1.Any help???
Without seeing a screenshot of your actual model, it's difficult to give specific detail, but here is some general advice that you may find helpful.
I would debug a Simulink model pretty much the same way that I would debug any other piece of code -- by trying to understand exactly what is happening in my model by examining the output at each step.
For starters, I would make use of the To Workspace block, to see what the data actually looks like at the intermediate stages of your model. This will help you to identify where things are going wrong. For example, what is the output of rgb2intensity? Is this function transforming your data in the way that you expect it to? Ask that question of each block in your model if you have to. If the block is not behaving the way that you expect it to, then you can ask why?
Another tool that might be useful for you is the Simulink Debugger.
so essentially I have the task to learn matlab decently well in the next few weeks, and I want to really be able to impress the people i'm learning it for, so if you guys have advice, I would greatly appreciate it.
What I'm dealing with Throughout the course of my job I will be dealing largely in two areas.
Formatting and importing data from excel
Interpreting the data in a meaningful way.
I realize the second category is extremely broad, but essentially i'm working with arrays of pricing information to set some standard price for a commodity, so the analysis that I will be doing will be less computationally heavy and will focus more on error getting the data in correctly and accurately and making sure all of the functions that I want to perform are executed correctly. Very basic stuff.
So here's what I'm looking for:
What are the most important topics so that I can import data from excel into matlab perfectly?
What should I study just to get basic functions down that will be applied to entire matrices/arrays?
What should I study prior to studying anything else to get a firm foundation in the subject? (So I don't make stupid syntax errors, etc.)?
Your help is greatly appreciated :)
The first two things you'll want to learn about are (1) the Matlab GUI and (2) the basics of Matlab syntax. A quick visit to the Getting Started section of the Matlab documentation will do you wonders. It should take you less than a day to get through the basics. Do note that the Matlab documentation is generally excellent, and you should use it often.
Beyond that, there are a number of questions here about importing Excel into Matlab (importing from excel, exporting to excel), and there's some excellent docs on the Mathworks site as well (xlsread, xlswrite). Read them and become familiar with the details and common problems. Good luck!
Since you will be using excel it is likely the data you import could come out in cell format. Look into the differences between indexing using matrices and cells. Also, learn early on the difference between [operator] and .[operator] for vector calculations. The second one is element-wise and will most likely solve a lot of issues you'll come across if your records are meant to only relate to one row/column of data
Perhaps you have gained the MATLAB skills you need now, but hopefully this will help someone following the same path later.
The Math Works (developers of MATLAB) run a site called MATLAB Cody.
This site offers a range of problems and a web based MATLAB interpreter so you can test your solutions. When you have a correct solution, you can compare with other solutions to the same problem. Then you can look up the functions others have used in the online MATLAB documentation for more understanding.
The problems focus on regular expressions and cell manipulation which will be very relevant to importing and manipulating Excel data.