README.md

Project: PyClustering

Version: 0.7.dev0

License: GNU General Public License

E-Mail: pyclustering@yandex.ru

PyClustering Wiki: https://github.com/annoviko/pyclustering/wiki

Documentation:

---

Proposals, questions, bugs:

In case of any questions, proposals or bugs related to the pyclustering please contact to pyclustering@yandex.ru or create an issue here.

---

PyClustering CI:

---

Based on:

• Python >= 3.4 windows 64-bit
• Python >= 3.4 linux 64-bit
• C++ 14 (MVS, GCC compilers)

---

Required packages:

• scipy, matplotlib, numpy, PIL

Index of packages for Windows:

• http://www.lfd.uci.edu/~gohlke/pythonlibs/

Index of packages for Linux:

• sudo apt-get install python3-numpy
• sudo apt-get install python3-scipy
• sudo apt-get install python3-matplotlib
• sudo apt-get install python3-pil

---

What is implemented in the project.

Clustering algorithms (module pyclustering.cluster):

• Agglomerative [Python, C++]
• BIRCH [Python]
• CLARANS [Python]
• CURE [Python, C++]
• DBSCAN [Python, C++]
• HSyncNet [Python, C++]
• K-Means [Python, C++]
• K-Medians [Python, C++]
• K-Medoids [Python, C++]
• OPTICS [Python, C++]
• ROCK [Python, C++]
• SOM-SC [Python, C++]
• SyncNet [Python, C++]
• Sync-SOM [Python]
• X-Means [Python, C++]

Oscillatory networks and neural networks (module pyclustering.nnet):

• CNN (Chaotic Neural Network) [Python]
• HHN (Oscillatory network based on Hodgkin-Huxley model) [Python]
• Hysteresis Oscillatory Network [Python]
• LEGION (Local Excitatory Global Inhibitory Oscillatory Network) [Python, C++]
• PCNN (Pulse-Coupled Neural Network) [Python, C++]
• SOM (Self-Organized Map) [Python, C++]
• Sync (Oscillatory network based on Kuramoto model) [Python, C++]
• SyncPR (Oscillatory network for pattern recognition) [Python, C++]
• SyncSegm (Oscillatory network for image segmentation) [Python, C++]

Graph Coloring Algorithms (module pyclustering.gcolor):

• DSatur [Python]
• Hysteresis [Python]
• GColorSync [Python]

Travelling Salesman Problem Algorithms (module pyclustering.tsp):

• AntColony [Python, C++]

Containers (module pyclustering.container):

• KD Tree [Python, C++]
• CF Tree [Python]

---

Examples:

The library contains examples for each algorithm and oscillatory network model:

Clustering examples: pyclustering/cluster/examples

Graph coloring examples: pyclustering/gcolor/examples

Oscillatory network examples: pyclustering/nnet/examples

---

Illustrations:

Cluster allocation on FCPS dataset collection by DBSCAN:

Cluster allocation by OPTICS using cluster-ordering diagram:

Image segmentation by Sync-SOM algorithm:

Partial synchronization (clustering) in Sync oscillatory network:

Cluster visualization by SOM (Self-Organized Feature Map)

---

Code examples:

Data clustering by CURE algorithm

from pyclustering.cluster import cluster_visualizer;
from pyclustering.cluster.cure import cure;

from pyclustering.utils import read_sample;

from pyclustering.samples.definitions import FCPS_SAMPLES;

# Input data in following format [ [0.1, 0.5], [0.3, 0.1], ... ].
input_data = read_sample(FCPS_SAMPLES.SAMPLE_LSUN);

# Allocate three clusters:
cure_instance = cure(input_data, 3);
cure_instance.process();
clusters = cure_instance.get_clusters();

# Visualize clusters:
visualizer = cluster_visualizer();
visualizer.append_clusters(clusters, None);
visualizer.show();

Data clustering by SYNC-SOM (bio-inspired) algorithm

from pyclustering.cluster import cluster_visualizer;
from pyclustering.cluster.syncsom import syncsom;

from pyclustering.samples.definitions import FCPS_SAMPLES;

from pyclustering.utils import read_sample, draw_dynamics;

# Input data in following format [ [0.1, 0.5], [0.3, 0.1], ... ].
input_data = read_sample(FCPS_SAMPLES.SAMPLE_TARGET);

# Create oscillatory network for cluster analysis
# where the first layer has size 9x9. Radius
# connectivity (similarity parameter) is 0.9.
# CCORE library (C/C++ part of the pyclustering library)
# is used to ensure high performance.
network = syncsom(input_data, 9, 9, 0.9, ccore = True);

# Simulate network (start processing) with collecting
# output dynamic.
(dyn_time, dyn_phase) = network.process(True, 0.999);

# Show structure of the first layer
network.show_som_layer();

# Show structure of the second layer
network.show_sync_layer();

# Show results of clustering
clusters = network.get_clusters();
visualizer = cluster_visualizer();
visualizer.append_clusters(clusters, input_data);
visualizer.show();

# Show output dynamic of the network (that is obtained
# from the second layer).
draw_dynamics(dyn_time, dyn_phase, x_title = "Time", y_title = "Phase", y_lim = [0, 2 * 3.14]);

Simulation of oscillatory network PCNN

from pyclustering.nnet.pcnn import pcnn_network, pcnn_visualizer;

# Create Pulse-Coupled neural network with 10 oscillators.
net = pcnn_network(10, ccore = ccore_flag);

# Perform simulation during 100 steps using binary external stimulus.
dynamic = net.simulate(100, [1, 1, 1, 0, 0, 0, 0, 1, 1, 1]);

# Allocate synchronous ensembles in the network.
ensembles = dynamic.allocate_sync_ensembles();

# Show output dynamic.
pcnn_visualizer.show_output_dynamic(dynamic); 

Simulation of chaotic neural network CNN

from pyclustering.samples.definitions import FCPS_SAMPLES;

from pyclustering.utils import read_sample;

from pyclustering.nnet.cnn import cnn_network, cnn_visualizer;

# load stimulus from file
stimulus = read_sample(SIMPLE_SAMPLES.SAMPLE_SIMPLE1);
        
# create chaotic neural network, amount of neurons should be equal to amout of stimulus
network_instance = cnn_network(len(stimulus));
        
# simulate it during 100 steps
output_dynamic = network_instance.simulate(steps, stimulus);
        
# display output dynamic of the network
cnn_visualizer.show_output_dynamic(output_dynamic);
        
# dysplay dynamic matrix and observation matrix to show clustering
# phenomenon.
cnn_visualizer.show_dynamic_matrix(output_dynamic);
cnn_visualizer.show_observation_matrix(output_dynamic);