To get the samples of the HDFS data:
To get other data:
- MNIST: Keras or scikit-learn library
- Twitter: https://www.kaggle.com/arkhoshghalb/twitter-sentiment-analysis-hatred-speech
- Adult: https://archive.ics.uci.edu/ml/datasets/Adult
For the full data, access must be requested:
To parse the HDFS data, use the Drain model from:
To reduce the HDFS data:
- Generate_HDFS_data.ipynb
The HDFS models are available under:
The PyTorch implementation of the DeepLog model:
To run experiments for Adult, Twitter and MNIST data:
- Robustness-{Adult,Twitter,MNIST}.ipynb
To visualize the results:
- Robustness-Plot.ipynb
Our results are available in the Robustness-{HDFS,Adult,Twitter,MNIST}.csv