Implement k-Means Clustering
Implement k-Means using the TensorFlow k-Means API. The TensorFlow API lets you scale k-means to large datasets by providing the following functionality:
- Clustering using mini-batches instead of the full dataset.
- Choosing more optimal initial clusters using k-means++, which results in faster convergence.
The TensorFlow k-Means API lets you choose either Euclidean distance or cosine distance as your similarity measure.