lib.sedna.core.lifelong_learning.lifelong_learning

Module Contents

Classes

LifelongLearning

Lifelong Learning (LL) is an advanced machine learning (ML) paradigm that

class lib.sedna.core.lifelong_learning.lifelong_learning.LifelongLearning(estimator, task_definition=None, task_relationship_discovery=None, task_mining=None, task_remodeling=None, inference_integrate=None, unseen_task_detect=None)

Bases: sedna.core.base.JobBase

Lifelong Learning (LL) is an advanced machine learning (ML) paradigm that learns continuously, accumulates the knowledge learned in the past, and uses/adapts it to help future learning and problem solving.

Sedna provide the related interfaces for application development.

Parameters

• estimator (Instance) – An instance with the high-level API that greatly simplifies machine learning programming. Estimators encapsulate training, evaluation, prediction, and exporting for your model.

• task_definition (Dict) – Divide multiple tasks based on data, see task_jobs.task_definition for more detail.

• task_relationship_discovery (Dict) – Discover relationships between all tasks, see task_jobs.task_relationship_discovery for more detail.

• task_mining (Dict) – Mining tasks of inference sample, see task_jobs.task_mining for more detail.

• task_remodeling (Dict) – Remodeling tasks based on their relationships, see task_jobs.task_remodeling for more detail.

• inference_integrate (Dict) – Integrate the inference results of all related tasks, see task_jobs.inference_integrate for more detail.

• unseen_task_detect (Dict) – unseen task detect algorithms with parameters which has registered to ClassFactory, see sedna.algorithms.unseen_task_detect for more detail

Examples

>>> estimator = XGBClassifier(objective="binary:logistic")
>>> task_definition = {
        "method": "TaskDefinitionByDataAttr",
        "param": {"attribute": ["season", "city"]}
    }
>>> task_relationship_discovery = {
        "method": "DefaultTaskRelationDiscover", "param": {}
    }
>>> task_mining = {
        "method": "TaskMiningByDataAttr",
        "param": {"attribute": ["season", "city"]}
    }
>>> task_remodeling = None
>>> inference_integrate = {
        "method": "DefaultInferenceIntegrate", "param": {}
    }
>>> unseen_task_detect = {
        "method": "TaskAttrFilter", "param": {}
    }
>>> ll_jobs = LifelongLearning(
        estimator=estimator,
        task_definition=task_definition,
        task_relationship_discovery=task_relationship_discovery,
        task_mining=task_mining,
        task_remodeling=task_remodeling,
        inference_integrate=inference_integrate,
        unseen_task_detect=unseen_task_detect
    )

train(self, train_data, valid_data=None, post_process=None, action='initial', **kwargs)

fit for update the knowledge based on training data.

Parameters

• train_data (BaseDataSource) – Train data, see sedna.datasources.BaseDataSource for more detail.

• valid_data (BaseDataSource) – Valid data, BaseDataSource or None.

• post_process (function) – function or a registered method, callback after estimator train.

• action (str) – update or initial the knowledge base

• kwargs (Dict) – parameters for estimator training, Like: early_stopping_rounds in Xgboost.XGBClassifier

Returns

train_history

Return type

object

update(self, train_data, valid_data=None, post_process=None, **kwargs)

evaluate(self, data, post_process=None, **kwargs)

evaluated the performance of each task from training, filter tasks based on the defined rules.

Parameters

• data (BaseDataSource) – valid data, see sedna.datasources.BaseDataSource for more detail.

• kwargs (Dict) – parameters for estimator evaluate, Like: ntree_limit in Xgboost.XGBClassifier

inference(self, data=None, post_process=None, **kwargs)

predict the result for input data based on training knowledge.

Parameters

• data (BaseDataSource) – inference sample, see sedna.datasources.BaseDataSource for more detail.

• post_process (function) – function or a registered method, effected after estimator prediction, like: label transform.

• kwargs (Dict) – parameters for estimator predict, Like: ntree_limit in Xgboost.XGBClassifier

Returns

• result (array_like) – results array, contain all inference results in each sample.

• is_unseen_task (bool) – true means detect an unseen task, false means not

• tasks (List) – tasks assigned to each sample.