Skip to content

stefanthaler/2017-fnlm-experiments-supervised

Repository files navigation

Towards a Neural Language Model for Signature Extraction

Description

This repository contains the source code and data on the experiments that where presented in the paper "Towards a Neural Language Model for Signature Extraction"

Preparation of experiment environment:

  • Requirements: python2.7, pip

Create virtual environment and activate it (optional)

  • pip install virtualenv
  • virtualenv exp
  • . exp/bin/activate.fish

Install dependencies

  • pip install keras==2.0.4
  • pip install scikit-learn==0.18.1
  • pip install tensorflow==1.1.0-cp27 (CPU version, GPU version is tensorflow-GPU)
  • pip install h5py==2.7.1 (for saving keras models)

Run the experiments

Experiments should be run in order. So to reproduce the neural language model experiment, run the script which name starts with 100, then 101, etc.

To evaluate:

Run the 900 script. The -e parameter specifies the directory where the results should be stored.

Citation

If you plan to use this work, please use the following citation

@inproceedings{thaler2017towards, title={Towards a neural language model for signature extraction from forensic logs}, author={Thaler, Stefan and Menkonvski, Vlado and Petkovic, Milan}, booktitle={Digital Forensic and Security (ISDFS), 2017 5th International Symposium on}, pages={1--6}, year={2017}, organization={IEEE} }

Work used in this paper:

IPLoM Implentation:

Paper of IPLoM:

Paper that provided IPLoM sourcecode:

LogCluster implementation

About

Experiments on neural language models for signature extraction

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published