Scribd
Upload
38 views1 page

LBG Algo

The LBG algorithm is a clustering algorithm similar to K-means that takes a set of input vectors and generates a subset of representative vectors as an output. It works by classifying vectors into K clusters based on similarity to cluster centers, updating the cluster centers based on the mean of each cluster, then repeating this process until the distortion between iterations falls below a threshold. The algorithm initializes a codebook, assigns training vectors to their closest codebook entry, recalculates the codebook based on cluster means, and iterates until convergence within a set number of steps or distortion threshold.

Uploaded by

Walid A. Sayssouk
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
38 views1 page

LBG Algo

The LBG algorithm is a clustering algorithm similar to K-means that takes a set of input vectors and generates a subset of representative vectors as an output. It works by classifying vectors into K clusters based on similarity to cluster centers, updating the cluster centers based on the mean of each cluster, then repeating this process until the distortion between iterations falls below a threshold. The algorithm initializes a codebook, assigns training vectors to their closest codebook entry, recalculates the codebook based on cluster means, and iterates until convergence within a set number of steps or distortion threshold.

Uploaded by

Walid A. Sayssouk
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

LBG Algorithm

LBG algorithm is like a K-means clustering algorithm which takes a set of input vectors S = {xi Rd | i = 1, 2, . . . , n} as input and generates a representative subset of vectors C = {cj Rd | j = 1, 2, . . . , K } with a user specied K << n as output according to the similarity measure. For the application of Vector Quantization (VQ), d = 16, K = 256 or 512 are commonly used. LBG Algorithm 1. Input training vectors S = {xi Rd | i = 1, 2, , n}. 2. Initiate a codebook C = {cj Rd | j = 1, 2, , K }. 3. Set D0 = 0 and let k = 0. 4. Classify the n training vectors into K clusters according to xi Sq if xi cq p xi cj p for j = q . 5. Update cluster centers cj , j = 1, 2, , K by cj = 6. Set k k + 1 and compute the distortion Dk =
1 | Sj | K j =1 xi Sj xi Sj

xi . xi cj p .

7. If (Dk1 Dk )/Dk > (a small number ), repeat steps 4 6. 8. Output the codebook C = {cj Rd | j = 1, 2, , K }, The convergence of LBG algorithm depends on the initial codebook C , the distortion Dk , and the threshold , in implementation, we need to provide a maximum number of iterations to guarantee the convergence.

You might also like