https://github.com/abdulfatir/subkmeans

Numpy and pyCUDA implementation of subKmeans
https://github.com/abdulfatir/subkmeans

clustering cuda kdd kmeans numpy pycuda python subspace-clustering

Last synced: 20 days ago
JSON representation

Numpy and pyCUDA implementation of subKmeans

• Host: GitHub
• URL: https://github.com/abdulfatir/subkmeans
• Owner: abdulfatir
• Created: 2018-03-06T15:02:08.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-04-20T11:12:09.000Z (over 6 years ago)
• Last Synced: 2024-10-07T04:41:14.211Z (about 1 month ago)
• Topics: clustering, cuda, kdd, kmeans, numpy, pycuda, python, subspace-clustering
• Language: Python
• Homepage:
• Size: 1.94 MB
• Stars: 1
• Watchers: 5
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Fast Subspace Clustering

## A GPU Implementation of *subKMeans*[1]

This is a CPU and GPU implementation of the KDD 2017 paper **Towards an Optimal Subspace for K-Means**[1]. 

The CPU implementation (`mode=cpu`) is written in Numpy. 

There are 2 GPU implementations:    

1. Using only PyCUDA and scikit-cuda API (`mode=gpu`)

2. Using PyCUDA with custom kernels optimized for this algorithm (`mode=gpu_custom`)

### Dependencies

* Numpy

* PyCUDA: `pip install pycuda`

* scikit-cuda: Install from source as described [here](https://github.com/lebedov/scikit-cuda/blob/master/docs/source/install.rst).

We tested using [commit #249538c](https://github.com/lebedov/scikit-cuda/commit/249538c95e68d891e2477d93eeff57941c99eb93).

* Matplotlib (for plots)

* scikit-learn (for computing NMI score)

Note: We tested this implementation only on Python 2. There are some issues with the GPU version on Python 3.

### Usage

Go to `src/`

```bash

python main.py -d= -k= -mode=

```

For help: `python main.py -h`

3 available modes: `cpu`, `gpu`, `gpu_custom`

#### Example Usage

`python main.py -d=wine -k=3 -mode=cpu`

##### Sample Output

```

[i] Itr 1: 24 points changed

[i] Itr 2: 7 points changed

[i] Itr 3: 7 points changed

[i] Itr 4: 2 points changed

[i] Itr 5: 1 points changed

[i] Itr 6: 0 points changed

[i] Results

[*] m: 2

[*] NMI: 0.87590

```

### References

[1] *Mautz et. al.* Towards an Optimal Subspace for K-Means