- Distribution of particles/jets raw features (see
eda/particle_eda.ipynb
andeda/jet_eda.ipynb
) - Distribution of particles preprocessed features
- 3D visualization of an event with two jets
- Figures representing the preprocessed features (powerpoint, tikz, other ?)
- delta angles
- projected momenta
- Figures representing the architecture of the neural networks
- Results of the neural networks
- ROC curves
- confusion matrices
- loss curves
- accuracy curves
- other ?
- ROOT to CSV conversion and feature extraction (see
preprocessing/make_dataset.py
) - Compute particle quantities w.r.t. jet quantities
- normalized energy $\hat{E}{p} = E_p / E j \in[0,,1]$ then scaled to $E=\hat{E}{p} - \hat{E}{\text{average}} \in [-\hat{E}{\text{average}},,1-\hat{E}{\text{average}}]\sim$ centered around 0
- momentum components w.r.t. jet direction
- delta angles between particle and jet
- other ?
- PCA / dim. reduction on the entire set of raw features to see what comes out
- What architectures?
- Train architectures on raw features
- Train architectures on preprocessed features
- How do we evaluate the performance of nns?
- Abstract (the last thing to do)
- Introduction (we begin with it and then redo it at the end)
- Related work → we do an overview of the different methods used in the literature
- Dataset → we describe the dataset
- Preprocessing → we describe the preprocessing steps and why we chose them
- Model → we describe the machine learning model(s) in great detail
- Results
- Conclusions
polar E = mass polar PX = pt polar PY = eta polar PZ = phi