Meta-Learning for Few-Shot Time Series Forecasting

This section of the README walks through how to train the models.

data_preprocessing.py + embedding.py

notes: The time-series data given in '/data/few_shot_data/...' already have done this step. For new raw time-series data, the two scripts can be used in this step.

main.py

Arguments help:

--baseNet: [mlp/cnn/lstm/cnnConlstm]
--dataset: the directory of saving pre-processed time-series data 
--update_step_target: update times of network
--fine_lr: leanring rate in this phase
--ppn: predict point number [10/20/30/40]
--device: [cpu/cuda]
--user_id: the name of the task that will be training, it can be found in ./config.py TRAINING_TASK_SET

training single task:

''' python main.py --baseNet [mlp/cnn/lstm/cnnConlstm] --dateset [few_shot_data/your defined data dir] --update_step_target 10 --fine_lr 0.001 --ppn [10/20/30/40] --device [cpu/cuda] --user_id 0001 '''

training all task:

''' python main.py --baseNet [mlp/cnn/lstm/cnnConlstm] --dateset few_shot_data --update_step_target 10 --fine_lr 0.001 --ppn [10/20/30/40] --device [cpu/cuda] '''

main.py

Argument help:

--maml: using 'maml mode' to training model
--update_step_train: the update times of baseNet on training-task set
--update_step_target: the update times of MetaNet on target task
--epoch: iteration times
--base_lr: the learning rate of baseNet
--meta_lr: the learning rate of MetaNet
--fine_lr: the learning rate of MetaNet during fine-tuing

training single task:

''' python main.py --baseNet [cnn/lstm/cnnConlstm] --maml --dataset few_shot_data --epoch 10 --update_step_train 10 --update_step_target 10 --base_lr 0.01 --meta_lr 0.01 --fine_lr 0.01 --ppn 10 --device [cpu/cuda] --user_id Wine '''

training all task:

''' python main.py --baseNet [cnn/lstm/cnnConlstm] --maml --dataset few_shot_data --epoch 10 --update_step_train 10 --update_step_target 10 --base_lr 0.01 --meta_lr 0.01 --fine_lr 0.01 --ppn 10 --device [cpu/cuda] '''