Bias and ReLU. One bias per vertex per filter.

Train various classifiers to get a baseline.

Return T_k X where T_k are the Chebyshev polynomials of order up to K. Complexity is O(KMN).

Filtering with Chebyshev interpolation Implementation: numpy. Data: x of size N x M x F N: number of sig

Coarsen a graph, represented by its adjacency matrix A, at multiple levels.

Return an approximation of the k-nearest cosine distances.

Compute exact pairwise distances.

Compute exact pairwise distances.

Embed the vocabulary using pre-trained vectors.

Generates len(factors) meshes, each of them is scaled by factors[i] and computes the transformations between them. Returns: M:

Return the embedding of a grid graph.

Explore the hyper-parameter space with an exhaustive grid search.

Keep in the vocaluary the M words who appear most often.

Given the graph Laplacian and a data matrix, return a data matrix which can be multiplied by the filter coefficients to filter X using the La

Return the Laplacian of the weigth matrix.

Upper-bound on the spectrum.

Adds to the inference model the layers required to generate loss.

Max pooling of size p. Should be a power of 2.

Normalize data to unit length.

Plot the spectrum of a list of multi-scale Laplacians L.

Remove a document if it contains less than nwords.

Replace randomly chosen edges by random edges.

Rescale the Laplacian eigenvalues in [-1,1].

ids: list of ids to play

Number of classes per document.

Number of documents per class.

Upsampling x