Official Implementation of KDD 2024 Paper Fast Unsupervised Deep Outlier Model Selection with Hypernetworks
To run the demo for HyPer, first install the required libraries by executing "pip install -r requirements.txt". It is working with Python 3.7+.
To run the demo for ADMoE on MLP, execute: "python 0_run_fast_demo.py".
More file description:
- HMLP.py, PE.py, and HNAEtrainer.py provide the implementation of the HN
- utils.py, init_utils.py, and deepsets.py includes a set of helper functions
- f-train provides the preloaded f-val
- datasets folder includes datasets