Error using bsxfun??

Hi, I want to display the results of the calculations that I made, I want to add two values with different matrix sizes, I use 2 ways by adding regular (+) and using bsxfun. But both errors. have code like below :

D = 25 ;
K = 11; 
h = 100; 
reg = 0.01; 
iteration=1000;  
 
w1 = 0.01*randn(D,h); 
w2 = 0.01*randn(h,K);
a = 0.5.*reg.*(W1.*W1);
b = 0.5.*reg.*(W2.*W2);
 
reg_los1 = a+b;
reg_los = bsxfun(@plus, a,b);
for i = 1:100;
    fprintf('Iteration: %d \n' , i);
end

If there are another way to fix, I hope someone can improve my code, thank you.

[Jan]

Please do not post just "But both errors", but pos the complete error messages in the forum. It is easier to fix a problem than to guess, what the problem is.

Answer 1

[madhan ravi]

Try :

D = 25 ;
K = 11; 
h = 100; 
reg = 0.01; 
iteration=1000;  % i can't see this being used anywhere?
 
w1 = 0.01*randn(D,h); 
w2 = 0.01*randn(h,K);
a = 0.5.*reg.*(w1.*w1); % -> matlab is case sensitive
b = 0.5.*reg.*(w2.*w2);
 
reg_los1 = a+b; %remove this line because im not sure this will work in previous versions
reg_los = bsxfun(@plus, a,b);
for i = 1:100;
    fprintf('Iteration: %d \n' , i);
end

[Oman Wisni]

Dear Mr.Ravi. I'm sorry. Because of my typo I forget one line code. Im trying your code but still error using bsxfun. Here I attached my full code :

D = 25 ;
K = 11; 
h = 100; 
reg = 0.01; 
data = rand(1,220); 
 
w1 = 0.01*randn(D,h); 
w2 = 0.01*randn(h,K);
a = 0.5.*reg.*(w1.*w1); % -> matlab is case sensitive
b = 0.5.*reg.*(w2.*w2);
 
reg_los = bsxfun(@plus, a,b);
loss = reg_los + data; % I adding my code
for i = 1:100;
    fprintf('Iteration: %d \n' ,'loss :%f \n' , i,loss);
end

[madhan ravi]

Did you see Jan’s comment?

[Oman Wisni]

The error like below :

Error using bsxfun
Non-singleton dimensions of the two input arrays must match each other.

Error in Training (line 49)
    reg_los = bsxfun(@plus, a,b);

.

[madhan ravi]

>> size(a)

ans =

    25   100

>> size(b)

ans =

   100    11
   
   

How can you add them?

[Oman Wisni]

yes, but I got the formula like that? so what I do?

[madhan ravi]

Better ask your teacher

[Oman Wisni]

yes, I will ask my teacher.

one question again before I ask my teacher. here I have code Python like below

a = rand (220,11);
b = np.argmax(a, axis = 1);
print 'accuracy :%.2f', (np.mean(b))

what is argmax and mean in matlab, how I create code above to matlab? Thanks

[madhan ravi]

please use google before asking , type argmax matlab and mean matlab you will find the answer in the first link

[Oman Wisni]

already asking google, but I don't understand. Yes I will try ask again.