pyclustering 0.8.0 library is collection of clustering algorithms, oscillatory networks, neural networks, etc.

GENERAL CHANGES:

• Optimization K-Means++ algorithm using numpy (pyclustering.cluster.center_initializer).
  See: no reference.

• Implemented K-Means++ initializer for CCORE (ccore.clst.kmeans_plus_plus).
  See: #382

• Optimization of X-Means clustering process by using KMeans++ for initial centers of split regions (pyclustering.cluster.xmeans, ccore.clst.xmeans).
  See: #382

• Implemented parallel Sync-family algorithms for C/C++ implementation (CCORE) only (ccore.sync).
  See: #170

• C/C++ implementation is used by default to increase performance.
  See: #393

• Ignore 'ccore' flag to use C/C++ if platform is not supported (pyclustering.core).
  See: #393

• Optimization of python implementation of the K-Means algorithm using numpy (pyclustering.cluster.kmeans).
  See: #403

• Implemented dynamic visualizer for oscillatory networks (pyclustering.nnet.dynamic_visualizer).
  See: no reference.

• Implemented C/C++ Hodgkin-Huxley oscillatory network for image segmentation in CCORE to increase performance (ccore.hhn, pyclustering.nnet.hhn).
  See: #217

• Performance optimization for CCORE on linux platform.
  See: no reference.

• 32-bit platform of CCORE is supported for Linux OS.
  See: #253

• 32-bit platform of CCORE is supported for Windows OS.
  See: #253

• Implemented method 'get_probabilities()' for obtaining belong probability in EM-algorithm (pyclustering.cluster.ema).
  See: #387

• Python implementation of CURE algorithm method 'get_clusters()' returns list of indexes (pyclustering.cluster.cure).
  See: #384

• Implemented parallel processing for X-Means algorithm (ccore.clst.xmeans).
  See: #372

• Implemented pool threads for parallel processing (ccore.parallel).
  See: #383

• Optimization of OPTICS algorithm using KD-tree for searching nearest neighbors (pyclustering.cluster.optics, ccore.optics).
  See: #370

• Optimization of DBSCAN algorithm using KD-tree for searching nearest neighbors (pyclustering.cluster.dbscan, ccore.dbscan).
  See: #369

CORRECTED MAJOR BUGS:

• Incorrect type of medoid's index in K-Medians algorithm in case of Python 2.x (pyclustering.cluster.kmedoids).
  See: #415

• Hanging of method 'find_node' in KD-tree if it does not contain node with specified point and payload (pyclustering.container.kdtree).
  See: no reference.

• Incorrect clustering by CURE algorithm in some cases when data have a lot of identical points (pyclustering.cluster.cure).
  See: #414

• Segmentation fault in CURE algorithm in some cases when data have a lot of identical points (ccore.clst.cure).
  See: no reference.

• Incorrect segmentation by Python version of syncsegm - oscillatory network based on sync for image segmentation (pyclustering.nnet.syncsegm).
  See: #409

• Zero value of sigma under logarithm function in Python version of pyclustering X-Means algorithm (pyclustering.cluster.xmeans).
  See: #407

• Amplitude threshold is ignored during synchronous ensembles allocation for amplitude output dynamic 'allocate_sync_ensembles' - affect HNN, LEGION (pyclustering.utils).
  See: no reference.

• Wrong indexes can be returned during synchronous ensembles allocation for amplitude output dynamic 'allocate_sync_ensembles' - affect HNN, LEGION (pyclustering.utils).
  See: no reference.

• Amount of allocated clusters can be differ from amount of centers in X-Means algorithm (ccore.clst.xmeans).
  See: #389

• Amount of allocated clusters can be bigger than kmax in X-Means algorithm (pyclustering.cluster.xmeans, ccore.clst.xmeans).
  See: #388

• Corrected bug with returned nullptr in method 'kdtree_searcher::find_nearest_node()' (ccore.container.kdtree).
  See: no reference.