Lightweight Top-Quark Jet Classification with Deep Learning

This project explores how deep learning can be used to identify hadronically decaying top quarks from low-level detector-like data. Using a compact jet-image representation and a lightweight CNN, the model learns to distinguish Top-Quark jets from QCD background jets, achieving:

• Validation Accuracy: ~86.8%
• ROC AUC: ~0.93
• CPU-friendly training (≤16 GB RAM)

1. Physics Background

In high-energy proton–proton collisions (e.g., at the LHC), a hadronically decaying top quark follows:

t → Wb → q q̄′ b

The resulting three quarks appear as a three-prong energy pattern inside a single, boosted jet.
Standard QCD jets, by contrast, typically show only one dominant core.

These structural differences are important for analyses involving boosted-object tagging, new-physics searches, and trigger-level selections. To capture this substructure, raw particle 4-momentum is projected onto a 40×20 η–ϕ grid, forming a calorimeter-like jet image. Even this simplified representation preserves energy concentration, jet width, and multi-prong patterns—features that CNNs learn effectively.

2. Dataset

This project uses the open jet dataset from CERN (Zenodo):
https://zenodo.org/records/2603256

Each jet has up to 200 particles, stored as four-momentum values:

E, px, py, pz

(Zero-padded if fewer than 200 constituents.)

A preprocessed sample of 90k jets used in this project is available here: Google Drive:
https://drive.google.com/file/d/1ISTa0HIJZT8hH_Zf0tRVVoHQ5Ci1QJmK/view?usp=sharing

To create the Parquet file yourself:

import pandas as pd

df = pd.read_hdf("train.h5", "table").sample(n=90000)
df.to_parquet("jets90000.parquet.gzip", compression="gzip")

3. CNN Architecture

A compact CNN was chosen to balance physics performance with computational constraints (CPU-only training).

┌──────────────────────┐      ┌────────────────────────┐     ┌──────────────────────┐
│    Data Pipeline     │────▶ │   Jet Image Builder    │────▶│    CNN Classifier   │
│                      │      │                        │     │                      │
│ • 4-Momentum Input   │      │ • 40×20 Grid Mapping   │     │ • Conv5×5 (32)       │
│ • Normalization      │      │ • Energy Projection    │     │ • Conv3×3 (64)       │
│ • Train/Val Split    │      │ • Per-Jet Scaling      │     │ • Conv3×3 (128)      │
└──────────────────────┘      └────────────────────────┘     │ • FC(128) + Dropout  │
             │                         │                     │ • Softmax (Top/QCD)  │
             ▼                         ▼                     └───────────┬──────────┘
      ┌────────────────┐      ┌────────────────────────┐                 │
      │ Visualization  │◀────│     Evaluation Tools   │─────────────────┘
      │                │      │                        │
      │ • Jet Images   │      │ • ROC / AUC            │
      │ • Avg Heatmaps │      │ • Confusion Matrix     │
      │ • Comparisons  │      │ • Score Distribution   │
      └────────────────┘      └────────────────────────┘

Key hyperparameters:

HYPERPARAMETER	VALUE	DESCRIPTION
Initial Learn Rate	5e-3	Reduces risk of overshooting the optimum
Learn Rate Schedule	Piecewise	Simple and stable decay strategy
Learn Rate Decay	0.3	Multiplies LR by 0.3 at each drop step
Learn Rate Drop Period	4 epochs	Decays LR every 4 epochs
L2 Regularization	1e-4	Light penalty to control overfitting
Batch Size	64	Balanced between training speed and RAM use
Max Epochs	12	Model begins to overfit past ~14 epochs
Optimizer	Adam	Stable and adaptive for small CNNs

4. Evaluation Results

Achieved 86.8% validation accuracy and AUC ≈ 0.93 using 90k training samples.

Confusion Matrix

ROC Curve

Score Distribution

5. Jet Image Visualizations

The project includes visualization tools that helped me understand the physics behind the classification:

Single Jet Visualization: A single jet image showing the sparse, tower-like energy pattern.

Signal vs Background: Average jet images highlighting structural differences between top and QCD jets.

Side-by-Side Comparisons: Direct comparison of individual signal and background jets.

6. Folder Structure

quark_detection/
│
├── data/
│   ├── jets90000.parquet.gzip        # Refer to gdrive
│   ├── cnn_v1_data.mat               # Train/val data
|   ├── cnn_v1_split.mat              # Split info
│   └── cnn_v1_eval.mat               # Evaluation outputs
|   
│
├── model/
│   └── cnn_model.mat                 # Final trained CNN model
│
├── results/
│   ├── v1_confusion_matrix.png
│   ├── v1_roc_curve.png
│   ├── v1_score_distribution.png
│   ├── jet123.png
│   ├── avg_quark.png
│   └── sig_vs_back.png
│
├── scripts/
|   ├── cnn.m
│   ├── evaluatem.m
│   └── visualize_jets.m
│
└── README.md

7. Running the Project

• Step 1 — Convert data to jet images
  run scripts/cnn_v1_data.m

• Step 2 — Train the CNN
  run scripts/cnn_v1_split.m

• Step 3 — Evaluate the model
  run scripts/cnn_v1_eval.m

• Step 4 — Visualize jets
  visualize_jets("single", 50);

8. Conclusion

This project demonstrates that meaningful jet substructure discrimination can be achieved using a lightweight deep learning approach that is accessible beyond large computing clusters. By operating on compact jet-image representations and a small convolutional neural network, the model achieves strong signal–background separation while remaining feasible to train on a standard laptop with modest memory requirements.

In addition to classification performance, the accompanying visualizations highlight the physically interpretable features learned by the network, such as the characteristic multi-prong structure of hadronically decaying top-quark jets. This combination of computational efficiency, interpretability, and competitive performance makes the approach suitable for exploratory studies, education, and rapid prototyping in high-energy physics analyses.