Skip to content

Latest commit

 

History

History
37 lines (22 loc) · 985 Bytes

README.md

File metadata and controls

37 lines (22 loc) · 985 Bytes
Raw

Image of Algorithm

Repository for all experiments performed in the area of detection, emulation and clustering of programs based on execution traces.

Experiments

1. Sorting

2. Detection

3. Classification

  • algorithms.py: implementation for IOAs
  • extractor.py: extracts features from IOAs
  • generator.py: generates IOAs and saves them to csv (any length)
  • process.py: prepares IOAs for training (standardizes length)
  • models.py: training and validation MLP, CNN & LSTM with IOAs
  • default_plot.py: plots training and validation results

4. Clustering

5. Emulation

Ideas from Notebook

First IOA

The ability to group programs based on their semantics.

Second IOA

The ability to interact with the programs learned from (to emulate/predict).

Similar to 2 Mentors

First one behaves like an oracle: only provides the right answers. Second one describes the thought process, so the observer can see how certain answers are arrived to.