Skip to content

ironfrown/ts_anomaly_detection_by_denoising

Repository files navigation

Anomaly detection in time series by data denoising

This project aims to apply various denoising techniques to identify anomalies in various forms of time series, e.g. in vibrations, medical or radio signals.

Status

Current version: V3.00

  • Several tests conducted on the full-QAE to determine its performance in response to the size of latent/trash space.

The current work is summarised here:

  • qiskit_qae/ts_qiskit_qae_binary_...: TS QAE based on binary encoding of TS windows developed and tested, abandoned.
  • qiskit_qae/ts_qiskit_qae_unary_...: TS QAE based on unary encoding of TS windows developed and tested, abandoned.
  • qiskit_qae/ts_qiskit_qae_angles_...: TS QAE based on angle encoding of TS windows developed and tested, results are promising.

Files

This repository consists of the following groups of notebooks:

  • dataset: samples of data used by the notebooks
  • images: images used in the notebooks
  • classic_pytorch: classically computed solutions with PyTorch
  • classic_tensorflow: classically computed solutions with Tensorflow
  • qiskit_qae: quantum autoencoders with Qiskit
  • pennylane_qae: quantuym autoencoders with PennyLane
  • runs: experimental runs with important results
  • tests: small tests of various features
  • tutorials: tutorials and demos from external sources
  • legacy: prgrams no longer needed or used

Software installation

To get installation details of Python, Anaconda / Miniconda, Pytorch, Tensorflow, Qiskit and PennyLane please refer to their respective web sites.

About

Anomaly detection in signals by denoising

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published