Kernel-Based Complexity Measure for the Study of Double Descent in Neural Networks

We investigate double descent phenomena in neural networks, using a novel kernel-based complexity measure. Building upon the work of Curth et al. 2023

Installation

1. Clone the repository:

   git clone https://github.com/VictorBaillet/double-descent
   cd double-descent

2. Set up the environment:

   python -m venv venv
   source venv/bin/activate  # For Windows, use `venv\Scripts\activate`
   pip install -r requirements.txt

3. Launch the Jupyter notebook for a hands-on experience with the results.

Run experiments

To run the experiments on your machine :

```bash
python main.py config/[experiment_name].json
```

Datasets Employed

The study employs two datasets :

• MNIST
• CIFAR-10