The distance between two layers.
The distance between the layers of two neural networks.
The distance between two skip-connections.
The distance between the skip-connections of two neural networks.
The distance between two neural networks. Args: x: An instance of NetworkDescriptor. y: An instance of NetworkDescriptor
Calculate the edit distance.
train_x: A list of neural architectures.
train_y: A list of neural architectures.
The Euclidean distance between two vectors.
Use Bourgain algorithm to embed the neural architectures based on their edit-distance.
- distance_matrix: A matrix of edit-distances.
Check if the target descriptor is in the descriptors.
Gaussian process regressor.
- alpha: A hyperparameter.
Fit the regressor with more data.
train_x: A list of NetworkDescriptor.
train_y: A list of metric values.
Incrementally fit the regressor.
Fit the regressor for the first time.
Predict the result.
- train_x: A list of NetworkDescriptor.
y_mean: The predicted mean.
y_std: The predicted standard deviation.
A Bayesian optimizer for neural architectures.
searcher: The Searcher which is calling the Bayesian optimizer.
t_min: The minimum temperature for simulated annealing.
metric: An instance of the Metric subclasses.
gpr: A GaussianProcessRegressor for bayesian optimization.
beta: The beta in acquisition function. (refer to our paper)
search_tree: The network morphism search tree.
Fit the optimizer with new architectures and performances.
x_queue: A list of NetworkDescriptor.
y_queue: A list of metric values.
Generate new architecture.
descriptors: All the searched neural architectures.
timeout: An integer. The time limit in seconds.
sync_message: the Queue for multiprocessing return value.
graph: An instance of Graph. A morphed neural network with weights.
father_id: The father node ID in the search tree.
Elements to be sorted according to metric value.
Elements to be reversely sorted according to metric value.
The network morphism search tree.
A recursive function to return the content of the tree in a dict.