Load cached dataset. :param training: Load training part if training is True, test part otherwise.

1D synthesis filter bank of an image tensor

1D analysis filter bank (along one dimension only) of an image Inputs: x (tensor): 4D input with the last two dimensions the spatial

Re-group scores respecting M4 rules. :param scores: Scores per group. :return: Grouped scores.

Prepares the filters to be of the right form for the afb2d function. In particular, makes the tensors the right shape. It takes mirror ima

Reflect the values in matrix *x* about the scalar values *minx* and *maxx*. Hence a vector *x* containing a long linearly increasing series is

Autocorrelation function. Args: ts: time series k: lag Returns: acf value

a/b where the resulted NaN or Inf are replaced by 0.

get modes on frequency domain: 'random' means sampling randomly; 'else' means sampling the lowest modes;

Function to do numpy like padding on tensors. Only works for 2-D padding. Inputs: x (tensor): tensor to pad pad (tuple)

Prepares the filters to be of the right form for the afb2d function. In particular, makes the tensors the right shape. It takes mirror ima

Prepares the filters to be of the right form for the sfb1d function. In particular, makes the tensors the right shape. It does not mirror i

Results visualization

1D analysis filter bank (along one dimension only) of an image without downsampling. Does the a trous algorithm. Inputs: x (tens

Forecastability Measure. Args: ts: time series Returns: 1 - the entropy of the fourier transformation of time series / entropy

The last window of insample size of all timeseries. This function does not support batching and does not reshuffle timeseries.

Prepares the filters to be of the right form for the afb2d_nonsep function. In particular, makes 2d point spread functions, and mirror imag

Prepares the filters to be of the right form for the sfb2d function. In particular, makes the tensors the right shape. It does not mirror

Prepares the filters to be of the right form for the sfb2d_nonsep function. In particular, makes 2d point spread functions. Does not mirror

Build mini-batch tensors from a list of (X, mask) tuples. Mask input. Create Args: data: len(batch_size) list of tuples (X, y).

Evaluate forecasts using M4 test dataset. :param forecast: Forecasts. Shape: timeseries, time. :return: sMAPE and OWA groupe