You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+7Lines changed: 7 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -133,6 +133,13 @@ You can set `custom_dist="cosine"` when initializing your `SpectralClusterer` ob
133
133
134
134
You can also use other distances supported by [scipy.spatial.distance](https://docs.scipy.org/doc/scipy/reference/spatial.distance.html), such as `"euclidean"` or `"mahalanobis"`.
135
135
136
+
### Affinity matrix
137
+
138
+
In our [ICASSP 2018 paper](https://google.github.io/speaker-id/publications/LstmDiarization/),
139
+
the affinity between two embeddings is defined as `(cos(x,y)+1)/2`.
140
+
141
+
You can also use other affinity functions by setting `affinity_function` when initializing your `SpectralClusterer` object.
142
+
136
143
### Auto-tune
137
144
138
145
We also support auto-tuning the `p_percentile` parameter of the `RowWiseThreshold` refinement operation, which was original proposed in [this paper](https://arxiv.org/abs/2003.02405).
0 commit comments