2 projects
tl4sm
This library performs transfer learning for time series classification between different data time series using a ConvLSTM2D auto-encoder architecture. Two transfer learning types are provided as options - fine-tuning a given number of layers, or re-using the pre-trained model weights as initialisation step. The output returns the f_score, accuracy, and training time.
tl4sm-generic
This library performs transfer learning for time series classification between different data time series using a user-specified keras machine learning model architecture. Two transfer learning types are provided as options - fine-tuning a given number of layers, or re-using the pre-trained model weights as initialisation step. The output returns the f_score, accuracy, and training time.
