Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Sign up
Appearance settings

🎉 New Accepted Solution: Top Quark Tagging Using Deep CNN #150

Discussion options

🎉 New Accepted Solution: Top Quark Tagging Using Deep CNN

Submitted by from .

Domain: High-Energy Physics
Method: Deep CNN
Result: >90% Accuracy

Project Description

tackled the challenge of discriminating top-quark jets from background by converting particle-jet constituent data into multi-channel "jet images" and training a custom convolutional neural network in MATLAB (via Deep Network Designer). The repository includes data preprocessing from the CERN Zenodo dataset, image-generation, model training, validation, and even hints at deployment workflows.

In the workflow, first ingested the raw four-vector data of up to 200 constituents per jet, padded where needed, then applied transformations into pixel-level channels (e.g., energy skewness, momentum kurtosis) and trained a deep CNN capable of distinguishing signal (top-quark) vs background jets. The model achieved over 90% test accuracy, demonstrating robust performance on this high-energy-physics classification task.

What’s especially strong is how structured the solution for reproducibility: clear folder layout (datasets, model, deploy), detailed README, explicit dataset source links, and MATLAB workflows. This makes it easy for others to follow, reproduce, or extend the work. Congratulations to for such a creative, well-documented, and technically rigorous contribution!

Important

Why it was accepted: Clear framing of the physics challenge, original use of CNN-based image modelling in MATLAB, reproducible code and data pipelines, and strong performance results on the target problem.

Tip

Technical highlights:

  • Custom deep convolutional blocks designed in MATLAB’s Deep Network Designer.
  • Multi-channel jet image generation from particle-level features (energy, momentum, pT skewness/kurtosis).
  • Training and validation workflows tailored to large-scale physics data (200 constituents / jet).
  • Achieved over 90% classification accuracy on test jets distinguishing top vs background.
  • Repository includes deploy folder hinting at hardware readiness or real-time processing ambitions.

Open in MATLAB Online
View Repository

At a glance

  • Challenge: Discriminate jets produced by top-quark decays from background jets in high-energy collisions.
  • Dataset/Models: CERN Zenodo dataset of jets (Energy, Momentum_X, Momentum_Y, Momentum_Z for up to 200 constituents per jet).
  • Toolboxes/Methods: MATLAB Deep Learning Toolbox, Deep Network Designer, custom CNN architecture, multi-channel jet-image preprocessing.
  • Key results: > 90% test classification accuracy distinguishing top-quark jets vs background.
  • Deployment: Repository includes "deploy" folder for future real-time or hardware-ready use (e.g., FPGA or embedded workflows).

Accepted to the MATLAB & Simulink Challenge Project Hub. Congratulations again to from for their remarkable contribution!

You must be logged in to vote

Replies: 0 comments

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
1 participant

AltStyle によって変換されたページ (->オリジナル) /