AutoTSFlow: An automated code-base for time-series classification

This repository contains the code for time-series (TS) classification with various state-of-the-art TS classification models. The entire pipeline is developed for easy integration of mlflow, hydra, and optuna sweeper, facilitating efficient experimentation, hyperparameter tuning, and configuration management.

The pipeline is developed in a modular way, where the models, datasets, and configurations can be easily added or modified.

1. The simplest way to run a model on a specific dataset is to run the following command in the terminal in the root directory of the repository:

python main.py

This will run a model on a dataset specified in the config file main_config.yaml, located in the config directory.

2. To optimize the hyperparameters of a model, run the following command in the terminal in the root directory of the repository:

python main.py --multirun

This will run a model on a dataset specified in the config file main_config.yaml, located in the config directory. However, this time, a search space specified in config/search_space/model_name will be used by optuna to find the optimal hyperparameters. A total number of trials is specified in the main_config.yaml file.

3. To run a model on a specific dataset, run the following command in the terminal, in the root directory of the repository:

python main.py "dataset_name=[Handwriting]"

For a multirun case:

python main.py --multirun "dataset_name=[Handwriting]"

To run for a specific model:

python main.py --multirun "dataset_name=[Handwriting]""models=LSTM_FCN"

Model name can be anything that is available in the codes/models directory, given corresponding configs are also available.

Similarly, other parameters can also be specified in the terminal, and passed as arguments.

Docker

To run the code in a docker container, run the following command in the terminal, in the root directory of the repository:

docker build -t ts_cl .
docker run -it ts_cl

This will build a docker image named ts_cl, and run a container with the image. The code can be run in the container as described above.

Mlruns

All the runs are stored in the mlruns directory. To visualize the runs, run the following command in the terminal, in the root directory of the repository:

mlflow ui

This will start a server, and the runs can be visualized in the browser at localhost:5000.

Requirements

environment.yaml file contains all the dependencies required to run the code. To install all the dependencies, run the following command in the terminal, given that anaconda is installed:

conda env create -f environment.yaml

This will create a conda environment named ts_cl with all the dependencies installed. It insall Pytorch with CPU support. To install Pytorch with GPU support, follow the instructions given here.

Datasets

This repository uses the datasets from the UEA & UCR Time Series Classification Repository. The datasets are automatically downloaded and stored in the data directory.

Models

We use the classification models available in tsai library. Models can be added to this repository by adding the corresponding config file in the config directory, and the corresponding model file in the codes/models directory.

Results

You can find the results in the following table. Each cell contains the accuracy of the corresponding model on the corresponding dataset. The results are obtained by running the models with the optimal hyperparameters found by optuna.

Dataset	GRU_FCN	InceptionTime	LSTM	LSTM_FCN
ECG200	0.91	0.91	0.82	0.92
HandMovementDirection	0.459459	nan	0.472973	0.486486
Handwriting	0.101176	0.0952941	0.0541176	0.0752941
ItalyPowerDemand	0.970845	0.969874	0.965986	0.910593

Authors

• Md Mijanur Rahman