def generalized_outer_product(mat): if len(mat.shape) == 1: return np.outer(mat, mat) elif len(mat.shape) == 2: return np.einsum('ij,ik->ijk', mat, mat) else: raise ArithmeticError
def predictions(weights, inputs): """weights is shape (num_weight_samples x num_weights) inputs is shape (num_datapoints x D)""" inputs = np.expand_dims(inputs, 0) for W, b in unpack_layers(weights): outputs = np.einsum('mnd,mdo->mno', inputs, W) + b inputs = nonlinearity(outputs) return outputs