from typing import Any, Dict, List
import keras
from aeon.classification.deep_learning import ResNetClassifier
from ml_grid.pipeline.data import pipe
from ml_grid.util.param_space import ParamSpace
[docs]
class ResNetClassifier_class:
"""A wrapper for the aeon ResNetClassifier time-series classifier.
This class provides a consistent interface for the ResNetClassifier,
including defining a hyperparameter search space.
Attributes:
algorithm_implementation: An instance of the aeon ResNetClassifier.
method_name (str): The name of the classifier method.
parameter_space (Dict[str, List[Any]]): The hyperparameter search space
for the classifier.
"""
[docs]
algorithm_implementation: ResNetClassifier
[docs]
parameter_space: Dict[str, List[Any]]
def __init__(self, ml_grid_object: pipe):
"""Initializes the ResNetClassifier_class.
Args:
ml_grid_object (pipe): An instance of the main data pipeline object.
"""
random_state_val = ml_grid_object.global_params.random_state_val
n_jobs_model_val = ml_grid_object.global_params.n_jobs_model_val
verbose_param = ml_grid_object.verbose
param_space = ParamSpace(
ml_grid_object.local_param_dict.get("param_space_size")
)
log_epoch = param_space.param_dict.get("log_epoch")
self.algorithm_implementation = ResNetClassifier()
self.method_name = "ResNetClassifier"
self.parameter_space = {
"n_residual_blocks": [
2,
3,
4,
], # The number of residual blocks of ResNet's model
"n_conv_per_residual_block": [
2,
3,
4,
], # The number of convolution blocks in each residual block
"n_filters": [
64,
128,
256,
], # The number of convolution filters for all the convolution layers in the same residual block
#'kernel_sizes': [3, 5, 7], # The kernel size of all the convolution layers in one residual block
"strides": [
1,
2,
], # The strides of convolution kernels in each of the convolution layers in one residual block
"dilation_rate": [
1,
2,
], # The dilation rate of the convolution layers in one residual block
"padding": [
"same",
"valid",
], # The type of padding used in the convolution layers in one residual block
"activation": [
"relu",
"tanh",
], # Keras activation used in the convolution layers in one residual block
"use_bias": [
True,
False,
], # Condition on whether or not to use bias values in the convolution layers in one residual block
"n_epochs": [log_epoch], # The number of epochs to train the model
"batch_size": [16, 32, 64], # The number of samples per gradient update
"use_mini_batch_size": [
True,
False,
], # Condition on using the mini-batch size formula Wang et al.
"callbacks": [None], # List of tf.keras.callbacks.Callback objects
#'file_path': ['./', './models/'], # File path when saving model_Checkpoint callback
"save_best_model": [False], # Whether or not to save the best model
"save_last_model": [False], # Whether or not to save the last model
#'best_file_name': ['best_model', 'best_classifier'], # Name of the file of the best model
#'last_file_name': ['last_model', 'last_classifier'], # Name of the file of the last model
"verbose": [verbose_param], # Whether to output extra information
"loss": [
"categorical_crossentropy"
], # Fit parameter for the keras model # 'mean_squared_error',
"optimizer": [
keras.optimizers.Adadelta(),
keras.optimizers.Adam(),
], # Optimizer for the model
"metrics": [["accuracy", "mae"]], # List of strings for metrics
}