Parametric UMAP changing n_components

@leriomaggio is already working on this.

Since Feb 16, 2021.

Hello,

I'm new with parametric UMAP, also i need to perform dimensionality reduction in a dataset and obtain 8 dimensional feature vector.

I understand that the parameter that i have to chage is n_components

I already change n_components from 2 to 8

# define the network
import tensorflow as tf
dims = (28, 28, 1)
n_components = 8
encoder = tf.keras.Sequential([
tf.keras.layers.InputLayer(input_shape=dims),
tf.keras.layers.Conv2D(
filters=32, kernel_size=3, strides=(2, 2), activation="relu", padding="same"
),
tf.keras.layers.Conv2D(
filters=64, kernel_size=3, strides=(2, 2), activation="relu", padding="same"
),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(units=256, activation="relu"),
tf.keras.layers.Dense(units=256, activation="relu"),
tf.keras.layers.Dense(units=n_components),
])
encoder.summary()

##############

but an error appears:

/usr/local/lib/python3.6/dist-packages/umap/umap_.py in fit(self, X, y)
2560 # Might be worth moving this into simplicial_set_embedding or _fit_embed_data
2561 disconnected_vertices = np.array(self.graph_.sum(axis=1)).flatten() == 0
-> 2562 self.embedding_[disconnected_vertices] = np.full(self.n_components, np.nan)
2563
2564 self.embedding_ = self.embedding_[inverse]

ValueError: shape mismatch: value array of shape (2,) could not be broadcast to indexing result of shape (0,8)

Please help me.

Thank you