| |$\mathbf {Input: }$| A query pattern x, a number of nearest neighbours k, a training set T. |
| |$\mathbf {Output:}$| A decision for x that class belong to. |
| |$\mathbf {Step:}$| |
| 1. Calculate the Euclidean distances of training samples in each class cl to x. |
| 2. Search the k-nearest centroid neighbours of x in each class cl, say |$T_{lk}^{\mathrm{ NCN}}(x)=\lbrace x_{lj}^{\mathrm{ NCN}}\in R^{d}\rbrace ^{k}_{j=1}$|. |
| 3. Compute the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| of the first j nearest neighbours of x using |$T^{\mathrm{ NCN}}_{lk}(x)$| and the corresponding distance |$d(x, \bar{u}_{lj}^{\mathrm{ NCN}}(x))$| between |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| and x. |
| 4. Allocate the weight |$\bar{W}_{lj}^{\mathrm{ NCN}}$| of the jth the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| in the set |$\bar{U}_{lk}^{\mathrm{ NCN}}(x)$|. |
| 5. Design pseudo-nearest centroid neighbour |$\bar{x}_{l}^{\mathrm{ PNCN}}(x)$| using |$\bar{W}_{lk}^{\mathrm{ NCN}}$| and |$d(x,\bar{u}_{lk}^{\mathrm{ NCN}}(x))$|. |
| 6. Assign the class c of the closest pseudo-nearest centroid neighbour to x. |
| |$\mathbf {Input: }$| A query pattern x, a number of nearest neighbours k, a training set T. |
| |$\mathbf {Output:}$| A decision for x that class belong to. |
| |$\mathbf {Step:}$| |
| 1. Calculate the Euclidean distances of training samples in each class cl to x. |
| 2. Search the k-nearest centroid neighbours of x in each class cl, say |$T_{lk}^{\mathrm{ NCN}}(x)=\lbrace x_{lj}^{\mathrm{ NCN}}\in R^{d}\rbrace ^{k}_{j=1}$|. |
| 3. Compute the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| of the first j nearest neighbours of x using |$T^{\mathrm{ NCN}}_{lk}(x)$| and the corresponding distance |$d(x, \bar{u}_{lj}^{\mathrm{ NCN}}(x))$| between |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| and x. |
| 4. Allocate the weight |$\bar{W}_{lj}^{\mathrm{ NCN}}$| of the jth the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| in the set |$\bar{U}_{lk}^{\mathrm{ NCN}}(x)$|. |
| 5. Design pseudo-nearest centroid neighbour |$\bar{x}_{l}^{\mathrm{ PNCN}}(x)$| using |$\bar{W}_{lk}^{\mathrm{ NCN}}$| and |$d(x,\bar{u}_{lk}^{\mathrm{ NCN}}(x))$|. |
| 6. Assign the class c of the closest pseudo-nearest centroid neighbour to x. |
| |$\mathbf {Input: }$| A query pattern x, a number of nearest neighbours k, a training set T. |
| |$\mathbf {Output:}$| A decision for x that class belong to. |
| |$\mathbf {Step:}$| |
| 1. Calculate the Euclidean distances of training samples in each class cl to x. |
| 2. Search the k-nearest centroid neighbours of x in each class cl, say |$T_{lk}^{\mathrm{ NCN}}(x)=\lbrace x_{lj}^{\mathrm{ NCN}}\in R^{d}\rbrace ^{k}_{j=1}$|. |
| 3. Compute the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| of the first j nearest neighbours of x using |$T^{\mathrm{ NCN}}_{lk}(x)$| and the corresponding distance |$d(x, \bar{u}_{lj}^{\mathrm{ NCN}}(x))$| between |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| and x. |
| 4. Allocate the weight |$\bar{W}_{lj}^{\mathrm{ NCN}}$| of the jth the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| in the set |$\bar{U}_{lk}^{\mathrm{ NCN}}(x)$|. |
| 5. Design pseudo-nearest centroid neighbour |$\bar{x}_{l}^{\mathrm{ PNCN}}(x)$| using |$\bar{W}_{lk}^{\mathrm{ NCN}}$| and |$d(x,\bar{u}_{lk}^{\mathrm{ NCN}}(x))$|. |
| 6. Assign the class c of the closest pseudo-nearest centroid neighbour to x. |
| |$\mathbf {Input: }$| A query pattern x, a number of nearest neighbours k, a training set T. |
| |$\mathbf {Output:}$| A decision for x that class belong to. |
| |$\mathbf {Step:}$| |
| 1. Calculate the Euclidean distances of training samples in each class cl to x. |
| 2. Search the k-nearest centroid neighbours of x in each class cl, say |$T_{lk}^{\mathrm{ NCN}}(x)=\lbrace x_{lj}^{\mathrm{ NCN}}\in R^{d}\rbrace ^{k}_{j=1}$|. |
| 3. Compute the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| of the first j nearest neighbours of x using |$T^{\mathrm{ NCN}}_{lk}(x)$| and the corresponding distance |$d(x, \bar{u}_{lj}^{\mathrm{ NCN}}(x))$| between |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| and x. |
| 4. Allocate the weight |$\bar{W}_{lj}^{\mathrm{ NCN}}$| of the jth the local mean vector |$\bar{u}_{lj}^{\mathrm{ NCN}}(x)$| in the set |$\bar{U}_{lk}^{\mathrm{ NCN}}(x)$|. |
| 5. Design pseudo-nearest centroid neighbour |$\bar{x}_{l}^{\mathrm{ PNCN}}(x)$| using |$\bar{W}_{lk}^{\mathrm{ NCN}}$| and |$d(x,\bar{u}_{lk}^{\mathrm{ NCN}}(x))$|. |
| 6. Assign the class c of the closest pseudo-nearest centroid neighbour to x. |
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.
