how to write code for given algorithm

Input:

train_data(d1,d2….dm) //training document vectors
with n features
test_data(d1,d2…dk) // testing document vectors p
features

Output:

sim_matrix // Similarity matrix

Steps:

1. [m,n]= size of train_data
2. [k,p]= size of test_data
3. Take Lamda=1 and sigma(1:n) =2
4. Take N*D=0 and NuD=0
5. for each document in test data and for each documentin training data
6. if test_data[r,j] . train_data[i,j]>0
7. then n1=0.5*(1+exp(-((train_data[r,j]-test_data[i,j])/sigma[j])^2))
8. if test_data[I,j] and train_data[r,j]=0
9. then n1=0
10. otherwise n1=-lamda
11. end if
12. N*D[i,r]=N*D[i,r]+n1[r,j]
13. end for
14. for each feature in train data
15. if train_data[r,j] and test[i,j]=0
16. then n2=0
17. otherwise n2=1
18. end if
19. NuD[i,r]=NuD[i,r]+n2[r,j]
20. end loop

[Image Analyst]

Read this http://www.mathworks.com/matlabcentral/answers/13205#answer_18099

Answer 1

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/8026-best-way-s-to-master-matlab