C++ API Reference for Intel® Data Analytics Acceleration Library 2020 Update 1
Contains classes for computing initial centroids for K-Means algorithm.
Namespaces | |
interface1 | |
Contains version 1.0 of the Intel(R) Data Analytics Acceleration Library (Intel(R) DAAL) interface. | |
Enumerations | |
enum | Method { deterministicDense = 0, defaultDense = 0, randomDense = 1, plusPlusDense = 2, parallelPlusDense = 3, deterministicCSR = 4, randomCSR = 5, plusPlusCSR = 6, parallelPlusCSR = 7 } |
enum | InputId { data } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm. More... | |
enum | DistributedStep2MasterInputId { partialResults } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm in the distributed processing mode. More... | |
enum | DistributedLocalPlusPlusInputDataId { internalInput = lastDistributedStep2MasterInputId + 1 } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm used with plusPlus and parallelPlus methods only on a local node. More... | |
enum | DistributedStep2LocalPlusPlusInputId { inputOfStep2 = lastDistributedLocalPlusPlusInputDataId + 1 } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm used with plusPlus and parallelPlus methods only on the 2nd step on a local node. More... | |
enum | DistributedStep3MasterPlusPlusInputId { inputOfStep3FromStep2 } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm used with plusPlus and parallelPlus methods only on the 3rd step on a master node. More... | |
enum | DistributedStep4LocalPlusPlusInputId { inputOfStep4FromStep3 = lastDistributedLocalPlusPlusInputDataId + 1 } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm used with plusPlus and parallelPlus methods only on a local node. More... | |
enum | DistributedStep5MasterPlusPlusInputId { inputCentroids, inputOfStep5FromStep2 } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm used with parallelPlus method only on a master node. More... | |
enum | DistributedStep5MasterPlusPlusInputDataId { inputOfStep5FromStep3 = lastDistributedStep5MasterPlusPlusInputId + 1 } |
Available identifiers of input objects for computing initial centroids for K-Means algorithm used with parallelPlus methods only on the 5th step on a master node. More... | |
enum | PartialResultId { partialCentroids, partialClusters = partialCentroids, partialClustersNumber } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode. More... | |
enum | DistributedStep2LocalPlusPlusPartialResultId { outputOfStep2ForStep3, outputOfStep2ForStep5 } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode used with plusPlus and parallelPlus methods only on the 2nd step on a local node. More... | |
enum | DistributedStep2LocalPlusPlusPartialResultDataId { internalResult = lastDistributedStep2LocalPlusPlusPartialResultId + 1 } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode used with plusPlus and parallelPlus methods only on the 2nd step on a local node. More... | |
enum | DistributedStep3MasterPlusPlusPartialResultId { outputOfStep3ForStep4 } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode used with plusPlus and parallelPlus methods only on the 3rd step on a master node. More... | |
enum | DistributedStep3MasterPlusPlusPartialResultDataId { rngState = lastDistributedStep3MasterPlusPlusPartialResultId + 1, outputOfStep3ForStep5 = rngState } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode used with parallelPlus method only on the 3rd step on a master node. More... | |
enum | DistributedStep4LocalPlusPlusPartialResultId { outputOfStep4 } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode used with plusPlus and parallelPlus methods only on the 4th step on a local node. More... | |
enum | DistributedStep5MasterPlusPlusPartialResultId { candidates, weights } |
Available identifiers of partial results of computing initial centroids for K-Means algorithm in the distributed processing mode used with parallelPlus method only on the 5th step on a master node. More... | |
enum | ResultId { centroids } |
Available identifiers of the results of computing initial centroids for K-Means algorithm. More... | |
enum InputId |
enum Method |
Available methods for computing initial centroids for K-Means algorithm
Enumerator | |
---|---|
deterministicDense |
Default: uses first nClusters points as initial centroids |
defaultDense |
Synonym of deterministicDense |
randomDense |
Uses random nClusters points as initial centroids |
plusPlusDense |
Kmeans++ algorithm by Arthur and Vassilvitskii (2007): http://ilpubs.stanford.edu:8090/778/1/2006-13.pdf [1] the first center is selected at random, each subsequent center is selected with a probability proportional to its contribution to the overall error |
parallelPlusDense |
Kmeans|| algorithm: scalable Kmeans++ by Bahmani et al. (2012) http://vldb.org/pvldb/vol5/p622_bahmanbahmani_vldb2012.pdf [2] |
deterministicCSR |
Uses first nClusters points as initial centroids for data in a CSR numeric table |
randomCSR |
Uses random nClusters points as initial centroids for data in a CSR numeric table |
plusPlusCSR |
Kmeans++ algorithm Arthur and Vassilvitskii (2007) http://ilpubs.stanford.edu:8090/778/1/2006-13.pdf [1] for the data in a CSR numeric table: the first center is selected at random, each subsequent center is selected with a probability proportional to its contribution to the overall error |
parallelPlusCSR |
Kmeans|| algorithm: scalable Kmeans++ by Bahmani et al. (2012) http://vldb.org/pvldb/vol5/p622_bahmanbahmani_vldb2012.pdf [2] for the data in a CSR numeric table |
enum PartialResultId |
Enumerator | |
---|---|
partialCentroids |
Table with the sum of observations assigned to centroids |
partialClusters |
Table with the sum of observations assigned to centroids
|
partialClustersNumber |
Table with the number of observations assigned to centroids
|
enum ResultId |
For more complete information about compiler optimizations, see our Optimization Notice.