TSFEDL

TSFEDL is a Python library that contains a wide range of models for working with time-series data using hybrid deep learning models that uses both a Convolutional Neural Network (CNN) and a Recurrent Neural Network (RNN). The use of these models allow both to capture the spatial features and the time dependencies between timesteps.

This library contains the implementation of 22 architectures both in Tensorflow+Keras and PyTorch.

Installation

To install TSFEDL, we recommend using pip.

pip install -e .

• Models Keras
  • OhShuLih()
  • KhanZulfiqar()
  • ZhengZhenyu()
  • HouBoroui()
  • WangKejun()
  • ChenChen()
  • KimTaeYoung()
  • GenMinxing()
  • FuJiangmeng()
  • ShiHaotian()
  • HuangMeiLing()
  • LihOhShu()
  • GaoJunLi()
  • WeiXiaoyan()
  • KongZhengmin()
  • YildirimOzal()
  • CaiWenjuan()
  • KimMinGu()
  • HtetMyetLynn()
  • ZhangJin()
  • YaoQihang()
  • YiboGao()
  • SharPar()
  • DaiXiLi()
• Models Pytorch
  • TSFEDL_BaseModule()
  • OhShuLih()
  • KhanZulfiqar()
  • ZhengZhenyu()
  • WangKejun()
  • ChenChen()
  • KimTaeYoung()
  • GenMinxing()
  • FuJiangmeng()
  • ShiHaotian()
  • HuangMeiLing()
  • LihOhShu()
  • GaoJunLi()
  • WeiXiaoyan()
  • KongZhengmin()
  • YildirimOzal()
  • CaiWenjuan()
  • HtetMyetLynn()
  • ZhangJin()
  • YaoQihang()
  • YiboGao()
  • SharPar()
  • DaiXiLi()
• PyTorch Blocks
  • ConvBlockYiboGao()
  • AttentionBranchYiboGao()
  • RTABlock()
  • SqueezeAndExcitationModule()
  • DenseNetTransitionBlock()
  • DenseNetConvBlock()
  • DenseNetDenseBlock()
  • SpatialAttentionBlockZhangJin()
  • TemporalAttentionBlockZhangJin()
• Keras Blocks
  • densenet_transition_block()
  • densenet_conv_block()
  • densenet_dense_block()
  • squeeze_excitation_module()
  • conv_block_YiboGao()
  • attention_branch_YiboGao()
  • RTA_block()
  • spatial_attention_block_ZhangJin()
  • temporal_attention_block_ZhangJin()
• Data Functions
  • get_mit_bih_segments()
  • read_mit_bih()
  • MIT_BIH()
• Utils Functions
  • TimeDistributed()
  • flip_indices_for_conv_to_lstm()
  • flip_indices_for_conv_to_lstm_reshape()
  • check_inputs()
  • full_convolution()

• Index

Release history

1.0.0 (1 Dec 2021)

Description

First release of the library.

1.0.1 (21 Jan 2022)

Description

Name of the library has changed and new models have been added.

• Added: HongTan model

• Changed: The library’s name is now TSFEDL.

• Deprecated:

• Removed:

• Fixed: Minor bugs.

1.0.2 (22 Mar 2022)

Description

Minor bugs are fixed.

• Added: Documentation of the base model in PyTorch and PyPi configuration files.

• Changed: Changed name of the base module in PyTorch for TSFEDL_BaseModule

• Deprecated:

• Removed:

• Fixed: Minor bugs.

1.0.3 (25 Mar 2022)

Description

Minor bugs are fixed, including a valid testing of all the networks.

• Added: Documentation and continuous integration with Travis.

• Changed:

• Deprecated:

• Removed:

• Fixed: Testing scripts.

1.0.5 (31 Mar 2022)

Description

Minor bugs are fixed, installation fixed for gpu servers.

• Added: Documentation and gpu testing.

• Changed:

• Deprecated:

• Removed:

• Fixed: Testing scripts and installation.

1.0.6 (5 Jul 2022)

Description

Minor bugs are fixed. Unused parameters removed.

• Added:

• Changed: Documentation

• Deprecated:

• Removed: Unused parameters

• Fixed: Testing scripts and installation.

1.0.7 (4 Sep 2022)

Description

New models added from 2021 and 2022

• Added: SharPar and DaiXiLi models

• Changed: Documentation

• Deprecated:

• Removed:

• Fixed: Testing scripts and installation.

1.0.7.1 (6 Feb 2023)

Description

Installation updated for tensorflow.

• Added:

• Changed:

• Deprecated:

• Removed:

• Fixed: Requirements and config to support only tensorflow as package name.