This example represents the output the t-SNE dimensionality reduction algorithm on embeddings computed from Unicode emojis using Keras (see Bradley Pallen's repository for more details). The example leverages the Labels element to visualize the Unicode emojis in the 2D coordinate system computed by the t-SNE algorithm.



In [ ]:

    
import pandas as pd
import holoviews as hv
hv.extension('bokeh')

Declaring data



In [ ]:

    
emoji_df = pd.read_csv('../../../assets/emoji_embeddings.csv', index_col=0)
emojis = hv.Labels(emoji_df, label='Emoji t-SNE Embeddings').redim.range(x=(-30, 20), y=(-20, 20))

Plot



In [ ]:

    
emojis.opts(width=1000, height=800, xaxis=None, yaxis=None)