migueldsw / ml-msc Goto Github PK

Compute the dissimilarity matrix of a given dataset.
Input: m instances (n-dimensional).
Output: Dissimilarity Matrix (mxm).
Adopt Euclidean distance.

Use the Multiple Features Data Set from UCI Repository (https://archive.ics.uci.edu/ml/datasets/Multiple+Features).
Import data into the project.

t <- 0
G = (g_1, ... , g_K) } vector of medoid vectors
g_k = random distinct medoid
/\ = (.\ _1 , ... , .\ _K ) } vector of relevance weight vectors
.\ _k = (1,...,1)
U = (u_1, ..., u_n ) } vector of membership degree vectors

u_t = J(v_t) = J(G, /\ , U)

"divide by zero" in eq. (6) -> getInitialVectorOfMembershipDegreeVectors(E,K)

RuntimeWarning: divide by zero encountered in double_scalars
u_i_k = ( argSum ** (1./(m-1.)) ) ** -1.

Include time cost evaluation in each step.

t=_ in _ secs

Run the algorithm, using the multiple features data set.
Adopting:
K = 10; m = 1.6 ; T = 150 ; e = 10¹⁰