DisjointCNNClassifier¶
- class DisjointCNNClassifier(n_layers=4, n_filters=64, kernel_size=None, dilation_rate=1, strides=1, padding='same', activation='elu', use_bias=True, kernel_initializer='he_uniform', pool_size=5, pool_strides=None, pool_padding='valid', hidden_fc_units=128, activation_fc='relu', n_epochs=2000, batch_size=16, use_mini_batch_size=False, random_state=None, verbose=False, loss='categorical_crossentropy', metrics='accuracy', optimizer=None, file_path='./', save_best_model=False, save_last_model=False, save_init_model=False, best_file_name='best_model', last_file_name='last_model', init_file_name='init_model', callbacks=None)[source]¶
Disjoint Convolutional Neural Netowkr classifier.
Adapted from the implementation used in [1].
- Parameters:
- n_layersint, default = 4
Number of 1+1D Convolution layers.
- n_filtersint or list of int, default = 64
Number of filters used in convolution layers. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- kernel_sizeint or list of int, default = [8, 5, 5, 3]
Size of convolution kernel. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- dilation_rateint or list of int, default = 1
The dilation rate for convolution. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- stridesint or list of int, default = 1
The strides of the convolution filter. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- paddingstr or list of str, default = “same”
The type of padding used for convolution. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- activationstr or list of str, default = “elu”
Activation used after the convolution. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- use_biasbool or list of bool, default = True
Whether or not ot use bias in convolution. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- kernel_initializer: str or list of str, default = “he_uniform”
The initialization method of convolution layers. If input is set to a list, the lenght should be the same as n_layers, if input is int the a list of the same element is created of length n_layers.
- pool_size: int, default = 5
The size of the one max pool layer at the end of the model, default = 5.
- pool_strides: int, default = None
The strides used for the one max pool layer at the end of the model, default = None.
- pool_padding: str, default = “valid”
The padding method for the one max pool layer at the end of the model, default = “valid”.
- hidden_fc_units: int, default = 128
The number of fully connected units.
- activation_fc: str, default = “relu”
The activation of the fully connected layer.
- n_epochsint, default = 2000
The number of epochs to train the model.
- batch_sizeint, default = 16
The number of samples per gradient update.
- use_mini_batch_sizebool, default = False
Whether or not to use the mini batch size formula.
- random_stateint, RandomState instance or None, default=None
If int, random_state is the seed used by the random number generator; If RandomState instance, random_state is the random number generator; If None, the random number generator is the RandomState instance used by np.random. Seeded random number generation can only be guaranteed on CPU processing, GPU processing will be non-deterministic.
- verboseboolean, default = False
Whether to output extra information.
- lossstr, default = “categorical_crossentropy”
The name of the keras training loss.
- metricsstr or list[str], default=”accuracy”
The evaluation metrics to use during training. If a single string metric is provided, it will be used as the only metric. If a list of metrics are provided, all will be used for evaluation.
- optimizerkeras.optimizer, default = tf.keras.optimizers.Adam()
The keras optimizer used for training.
- file_pathstr, default = “./”
File path to save best model.
- save_best_modelbool, default = False
Whether or not to save the best model, if the modelcheckpoint callback is used by default, this condition, if True, will prevent the automatic deletion of the best saved model from file and the user can choose the file name.
- save_last_modelbool, default = False
Whether or not to save the last model, last epoch trained, using the base class method save_last_model_to_file.
- save_init_modelbool, default = False
Whether to save the initialization of the model.
- best_file_namestr, default = “best_model”
The name of the file of the best model, if save_best_model is set to False, this parameter is discarded.
- last_file_namestr, default = “last_model”
The name of the file of the last model, if save_last_model is set to False, this parameter is discarded.
- init_file_namestr, default = “init_model”
The name of the file of the init model, if save_init_model is set to False, this parameter is discarded.
- callbackskeras callback or list of callbacks,
default = None The default list of callbacks are set to ModelCheckpoint and ReduceLROnPlateau.
Notes
Adapted from the implementation from: https://github.com/Navidfoumani/Disjoint-CNN
References
[1]Foumani, Seyed Navid Mohammadi, Chang Wei Tan, and Mahsa Salehi.
“Disjoint-cnn for multivariate time series classification.” 2021 International Conference on Data Mining Workshops (ICDMW). IEEE, 2021.
Examples
>>> from aeon.classification.deep_learning import DisjointCNNClassifier >>> from aeon.datasets import load_unit_test >>> X_train, y_train = load_unit_test(split="train") >>> X_test, y_test = load_unit_test(split="test") >>> disjoint_cnn = DisjointCNNClassifier(n_epochs=20, ... batch_size=4) >>> disjoint_cnn.fit(X_train, y_train) DisjointCNNClassifier(...)
Methods
build_model
(input_shape, n_classes, **kwargs)Construct a compiled, un-trained, keras model that is ready for training.
clone
([random_state])Obtain a clone of the object with the same hyperparameters.
Convert y to required Keras format.
fit
(X, y)Fit time series classifier to training data.
fit_predict
(X, y, **kwargs)Fits the classifier and predicts class labels for X.
fit_predict_proba
(X, y, **kwargs)Fits the classifier and predicts class label probabilities for X.
get_class_tag
(tag_name[, raise_error, ...])Get tag value from estimator class (only class tags).
Get class tags from estimator class and all its parent classes.
get_fitted_params
([deep])Get fitted parameters.
Sklearn metadata routing.
get_params
([deep])Get parameters for this estimator.
get_tag
(tag_name[, raise_error, ...])Get tag value from estimator class.
get_tags
()Get tags from estimator.
load_model
(model_path, classes)Load a pre-trained keras model instead of fitting.
predict
(X)Predicts class labels for time series in X.
Predicts class label probabilities for time series in X.
reset
([keep])Reset the object to a clean post-init state.
save_last_model_to_file
([file_path])Save the last epoch of the trained deep learning model.
score
(X, y[, metric, use_proba, metric_params])Scores predicted labels against ground truth labels on X.
set_params
(**params)Set the parameters of this estimator.
set_tags
(**tag_dict)Set dynamic tags to given values.
summary
()Summary function to return the losses/metrics for model fit.
- build_model(input_shape, n_classes, **kwargs)[source]¶
Construct a compiled, un-trained, keras model that is ready for training.
In aeon, time series are stored in numpy arrays of shape (d,m), where d is the number of dimensions, m is the series length. Keras/tensorflow assume data is in shape (m,d). This method also assumes (m,d). Transpose should happen in fit.
- Parameters:
- input_shapetuple
The shape of the data fed into the input layer, should be (m, d).
- n_classesint
The number of classes, which becomes the size of the output layer.
- Returns:
- outputa compiled Keras Model
- clone(random_state=None)[source]¶
Obtain a clone of the object with the same hyperparameters.
A clone is a different object without shared references, in post-init state. This function is equivalent to returning
sklearn.clone
of self. Equal in value totype(self)(**self.get_params(deep=False))
.- Parameters:
- random_stateint, RandomState instance, or None, default=None
Sets the random state of the clone. If None, the random state is not set. If int, random_state is the seed used by the random number generator. If RandomState instance, random_state is the random number generator.
- Returns:
- estimatorobject
Instance of
type(self)
, clone of self (see above)
- fit(X, y) BaseCollectionEstimator [source]¶
Fit time series classifier to training data.
- Parameters:
- Xnp.ndarray or list
Input data, any number of channels, equal length series of shape
( n_cases, n_channels, n_timepoints)
or 2D np.array (univariate, equal length series) of shape(n_cases, n_timepoints)
or list of numpy arrays (any number of channels, unequal length series) of shape[n_cases]
, 2D np.array(n_channels, n_timepoints_i)
, wheren_timepoints_i
is length of seriesi
. Other types are allowed and converted into one of the above.Different estimators have different capabilities to handle different types of input. If
self.get_tag("capability:multivariate")
is False, they cannot handle multivariate series, so eithern_channels == 1
is true or X is 2D of shape(n_cases, n_timepoints)
. Ifself.get_tag( "capability:unequal_length")
is False, they cannot handle unequal length input. In both situations, aValueError
is raised if X has a characteristic that the estimator does not have the capability for is passed.- ynp.ndarray
1D np.array of float or str, of shape
(n_cases)
- class labels (ground truth) for fitting indices corresponding to instance indices in X.
- Returns:
- selfBaseClassifier
Reference to self.
Notes
Changes state by creating a fitted model that updates attributes ending in “_” and sets is_fitted flag to True.
- fit_predict(X, y, **kwargs) ndarray [source]¶
Fits the classifier and predicts class labels for X.
fit_predict produces prediction estimates using just the train data. By default, this is through 10x cross validation, although some estimators may utilise specialist techniques such as out-of-bag estimates or leave-one-out cross-validation.
Classifiers which override _fit_predict will have the
capability:train_estimate
tag set to True.Generally, this will not be the same as fitting on the whole train data then making train predictions. To do this, you should call fit(X,y).predict(X)
- Parameters:
- Xnp.ndarray or list
Input data, any number of channels, equal length series of shape
( n_cases, n_channels, n_timepoints)
or 2D np.array (univariate, equal length series) of shape(n_cases, n_timepoints)
or list of numpy arrays (any number of channels, unequal length series) of shape[n_cases]
, 2D np.array(n_channels, n_timepoints_i)
, wheren_timepoints_i
is length of seriesi
. other types are allowed and converted into one of the above.Different estimators have different capabilities to handle different types of input. If
self.get_tag("capability:multivariate")
is False, they cannot handle multivariate series, so eithern_channels == 1
is true or X is 2D of shape(n_cases, n_timepoints)
. Ifself.get_tag( "capability:unequal_length")
is False, they cannot handle unequal length input. In both situations, aValueError
is raised if X has a characteristic that the estimator does not have the capability for is passed.- ynp.ndarray
1D np.array of float or str, of shape
(n_cases)
- class labels (ground truth) for fitting indices corresponding to instance indices in X.- kwargsdict
key word arguments to configure the default cross validation if the base class default fit_predict is used (i.e. if function
_fit_predict
is not overridden. If_fit_predict
is overridden, kwargs may not function as expected. If_fit_predict
is not overridden, valid input iscv_size
integer, which is the number of cross validation folds to use to estimate train data. Ifcv_size
is not passed, the default is 10. Ifcv_size
is greater than the minimum number of samples in any class, it is set to this minimum.
- Returns:
- predictionsnp.ndarray
shape
[n_cases]
- predicted class labels indices correspond to instance indices in
- fit_predict_proba(X, y, **kwargs) ndarray [source]¶
Fits the classifier and predicts class label probabilities for X.
fit_predict_proba produces probability estimates using just the train data. By default, this is through 10x cross validation, although some estimators may utilise specialist techniques such as out-of-bag estimates or leave-one-out cross-validation.
Classifiers which override _fit_predict_proba will have the
capability:train_estimate
tag set to True.Generally, this will not be the same as fitting on the whole train data then making train predictions. To do this, you should call fit(X,y).predict_proba(X)
- Parameters:
- Xnp.ndarray or list
Input data, any number of channels, equal length series of shape
( n_cases, n_channels, n_timepoints)
or 2D np.array (univariate, equal length series) of shape(n_cases, n_timepoints)
or list of numpy arrays (any number of channels, unequal length series) of shape[n_cases]
, 2D np.array(n_channels, n_timepoints_i)
, wheren_timepoints_i
is length of seriesi
. other types are allowed and converted into one of the above.Different estimators have different capabilities to handle different types of input. If
self.get_tag("capability:multivariate")
is False, they cannot handle multivariate series, so eithern_channels == 1
is true or X is 2D of shape(n_cases, n_timepoints)
. Ifself.get_tag( "capability:unequal_length")
is False, they cannot handle unequal length input. In both situations, aValueError
is raised if X has a characteristic that the estimator does not have the capability for is passed.- ynp.ndarray
1D np.array of float or str, of shape
(n_cases)
- class labels (ground truth) for fitting indices corresponding to instance indices in X.- kwargsdict
key word arguments to configure the default cross validation if the base class default fit_predict is used (i.e. if function
_fit_predict
is not overridden. If_fit_predict
is overridden, kwargs may not function as expected. If_fit_predict
is not overridden, valid input iscv_size
integer, which is the number of cross validation folds to use to estimate train data. Ifcv_size
is not passed, the default is 10. Ifcv_size
is greater than the minimum number of samples in any class, it is set to this minimum.
- Returns:
- probabilitiesnp.ndarray
2D array of shape
(n_cases, n_classes)
- predicted class probabilities First dimension indices correspond to instance indices in X, second dimension indices correspond to class labels, (i, j)-th entry is estimated probability that i-th instance is of class j
- classmethod get_class_tag(tag_name, raise_error=True, tag_value_default=None)[source]¶
Get tag value from estimator class (only class tags).
- Parameters:
- tag_namestr
Name of tag value.
- raise_errorbool, default=True
Whether a ValueError is raised when the tag is not found.
- tag_value_defaultany type, default=None
Default/fallback value if tag is not found and error is not raised.
- Returns:
- tag_value
Value of the
tag_name
tag in cls. If not found, returns an error ifraise_error
is True, otherwise it returnstag_value_default
.
- Raises:
- ValueError
if
raise_error
is True andtag_name
is not inself.get_tags().keys()
Examples
>>> from aeon.classification import DummyClassifier >>> DummyClassifier.get_class_tag("capability:multivariate") True
- classmethod get_class_tags()[source]¶
Get class tags from estimator class and all its parent classes.
- Returns:
- collected_tagsdict
Dictionary of tag name and tag value pairs. Collected from
_tags
class attribute via nested inheritance. These are not overridden by dynamic tags set byset_tags
or class__init__
calls.
- get_fitted_params(deep=True)[source]¶
Get fitted parameters.
- State required:
Requires state to be “fitted”.
- Parameters:
- deepbool, default=True
If True, will return the fitted parameters for this estimator and contained subobjects that are estimators.
- Returns:
- fitted_paramsdict
Fitted parameter names mapped to their values.
- get_params(deep=True)[source]¶
Get parameters for this estimator.
- Parameters:
- deepbool, default=True
If True, will return the parameters for this estimator and contained subobjects that are estimators.
- Returns:
- paramsdict
Parameter names mapped to their values.
- get_tag(tag_name, raise_error=True, tag_value_default=None)[source]¶
Get tag value from estimator class.
Includes dynamic and overridden tags.
- Parameters:
- tag_namestr
Name of tag to be retrieved.
- raise_errorbool, default=True
Whether a ValueError is raised when the tag is not found.
- tag_value_defaultany type, default=None
Default/fallback value if tag is not found and error is not raised.
- Returns:
- tag_value
Value of the
tag_name
tag in self. If not found, returns an error ifraise_error
is True, otherwise it returnstag_value_default
.
- Raises:
- ValueError
if raise_error is
True
andtag_name
is not inself.get_tags().keys()
Examples
>>> from aeon.classification import DummyClassifier >>> d = DummyClassifier() >>> d.get_tag("capability:multivariate") True
- get_tags()[source]¶
Get tags from estimator.
Includes dynamic and overridden tags.
- Returns:
- collected_tagsdict
Dictionary of tag name and tag value pairs. Collected from
_tags
class attribute via nested inheritance and then any overridden and new tags from__init__
orset_tags
.
- load_model(model_path, classes)[source]¶
Load a pre-trained keras model instead of fitting.
When calling this function, all functionalities can be used such as predict, predict_proba etc. with the loaded model.
- Parameters:
- model_pathstr (path including model name and extension)
The directory where the model will be saved including the model name with a “.keras” extension. Example: model_path=”path/to/file/best_model.keras”
- classesnp.ndarray
The set of unique classes the pre-trained loaded model is trained to predict during the classification task.
- Returns:
- None
- predict(X) ndarray [source]¶
Predicts class labels for time series in X.
- Parameters:
- Xnp.ndarray or list
Input data, any number of channels, equal length series of shape
( n_cases, n_channels, n_timepoints)
or 2D np.array (univariate, equal length series) of shape(n_cases, n_timepoints)
or list of numpy arrays (any number of channels, unequal length series) of shape[n_cases]
, 2D np.array(n_channels, n_timepoints_i)
, wheren_timepoints_i
is length of seriesi
other types are allowed and converted into one of the above.Different estimators have different capabilities to handle different types of input. If
self.get_tag("capability:multivariate")
is False, they cannot handle multivariate series, so eithern_channels == 1
is true or X is 2D of shape(n_cases, n_timepoints)
. Ifself.get_tag( "capability:unequal_length")
is False, they cannot handle unequal length input. In both situations, aValueError
is raised if X has a characteristic that the estimator does not have the capability for is passed.
- Returns:
- predictionsnp.ndarray
1D np.array of float, of shape (n_cases) - predicted class labels indices correspond to instance indices in X
- predict_proba(X) ndarray [source]¶
Predicts class label probabilities for time series in X.
- Parameters:
- Xnp.ndarray or list
Input data, any number of channels, equal length series of shape
( n_cases, n_channels, n_timepoints)
or 2D np.array (univariate, equal length series) of shape(n_cases, n_timepoints)
or list of numpy arrays (any number of channels, unequal length series) of shape[n_cases]
, 2D np.array(n_channels, n_timepoints_i)
, wheren_timepoints_i
is length of seriesi
. other types are allowed and converted into one of the above.Different estimators have different capabilities to handle different types of input. If
self.get_tag("capability:multivariate")
is False, they cannot handle multivariate series, so eithern_channels == 1
is true or X is 2D of shape(n_cases, n_timepoints)
. Ifself.get_tag( "capability:unequal_length")
is False, they cannot handle unequal length input. In both situations, aValueError
is raised if X has a characteristic that the estimator does not have the capability for is passed.
- Returns:
- probabilitiesnp.ndarray
2D array of shape
(n_cases, n_classes)
- predicted class probabilities First dimension indices correspond to instance indices in X, second dimension indices correspond to class labels, (i, j)-th entry is estimated probability that i-th instance is of class j
- reset(keep=None)[source]¶
Reset the object to a clean post-init state.
After a
self.reset()
call, self is equal or similar in value totype(self)(**self.get_params(deep=False))
, assuming no other attributes were kept usingkeep
.- Detailed behaviour:
- removes any object attributes, except:
hyper-parameters (arguments of
__init__
) object attributes containing double-underscores, i.e., the string “__”
runs
__init__
with current values of hyperparameters (result ofget_params
)- Not affected by the reset are:
object attributes containing double-underscores class and object methods, class attributes any attributes specified in the
keep
argument
- Parameters:
- keepNone, str, or list of str, default=None
If None, all attributes are removed except hyperparameters. If str, only the attribute with this name is kept. If list of str, only the attributes with these names are kept.
- Returns:
- selfobject
Reference to self.
- save_last_model_to_file(file_path='./')[source]¶
Save the last epoch of the trained deep learning model.
- Parameters:
- file_pathstr, default = “./”
The directory where the model will be saved
- Returns:
- None
- score(X, y, metric='accuracy', use_proba=False, metric_params=None) float [source]¶
Scores predicted labels against ground truth labels on X.
- Parameters:
- Xnp.ndarray or list
Input data, any number of channels, equal length series of shape
( n_cases, n_channels, n_timepoints)
or 2D np.array (univariate, equal length series) of shape(n_cases, n_timepoints)
or list of numpy arrays (any number of channels, unequal length series) of shape[n_cases]
, 2D np.array(n_channels, n_timepoints_i)
, wheren_timepoints_i
is length of seriesi
. other types are allowed and converted into one of the above.Different estimators have different capabilities to handle different types of input. If
self.get_tag("capability:multivariate")
is False, they cannot handle multivariate series, so eithern_channels == 1
is true or X is 2D of shape(n_cases, n_timepoints)
. Ifself.get_tag( "capability:unequal_length")
is False, they cannot handle unequal length input. In both situations, aValueError
is raised if X has a characteristic that the estimator does not have the capability for is passed.- ynp.ndarray
1D np.array of float or str, of shape
(n_cases)
- class labels (ground truth) for fitting indices corresponding to instance indices in X.- metricUnion[str, callable], default=”accuracy”,
Defines the scoring metric to test the fit of the model. For supported strings arguments, check sklearn.metrics.get_scorer_names.
- use_probabool, default=False,
Argument to check if scorer works on probability estimates or not.
- metric_paramsdict, default=None,
Contains parameters to be passed to the scoring function. If None, no parameters are passed.
- Returns:
- scorefloat
Accuracy score of predict(X) vs y.
- set_params(**params)[source]¶
Set the parameters of this estimator.
The method works on simple estimators as well as on nested objects (such as
Pipeline
). The latter have parameters of the form<component>__<parameter>
so that it’s possible to update each component of a nested object.- Parameters:
- **paramsdict
Estimator parameters.
- Returns:
- selfestimator instance
Estimator instance.