Configure the experimental environment and tools listed below:
pip3 install -r require.txt
Download UEA datasets
URL: http://timeseriesclassification.com/dataset.php
When obtained the UEA zip file, one should unzipthis file and add to:
repo: ./data/
see: ./processing/data.py
####configure introduce:
####dataset repo
save_file = "/mnt/54b93d63-868e-4b2c-aa09-c9109c3e67be/multivariateTSC/UCE"
#### GPU index used
GPU_number = 0
#### dataset name
dataset_name = "Beef"
### coeffecient for loss function
loss_coefficient = 0.1
### a scalarable temporate for distillation
temperate = 1.0
##### the weight of dual non-target class loss
beta = 1.0
##### the weight of dual target class loss
alpha = 1.0
###### learning rate
learning_rate = 0.0001
##### weight_decay value
weight_decay = 0.0005
#### training epoch
epoch = 200
#### save path
save_path = "save_model/"
python3 main.py -n 1 -b 1 -r 0.01 -s 50 -p 1 -e 500