← Latest papers
⚛️ phenomenology

Study of enhancing the signal-background separation for the Higgs boson production in association with a single top-quark in $pp$ collisions at the LHC

This paper presents a deep fully connected neural network with a masking mechanism to enhance signal-background separation for the rare Higgs boson production in association with a single top quark (tH) at the LHC, demonstrating superior statistical significance compared to conventional cut-based methods using simulated data from Runs 1, 2, and 3.

Original author: Rustam Azimzade, Alina Dadashova, Igor Boiko, Vladimir Lyubushkin, Nazim Huseynov

Published 2026-11-04
📖 5 min read🧠 Deep dive

Original author: Rustam Azimzade, Alina Dadashova, Igor Boiko, Vladimir Lyubushkin, Nazim Huseynov

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

The universe is built from a small set of fundamental particles that interact through invisible forces, a framework physicists call the Standard Model. At the heart of this theory sits the Higgs boson, a particle discovered in 2012 that explains how other particles acquire mass. While scientists have confirmed its existence, they still need to measure its properties with extreme precision to see if the current theory holds up or if there are cracks pointing toward new, undiscovered physics. One of the most difficult ways to study the Higgs boson is to watch it appear alongside a single top quark, the heaviest known elementary particle. This specific combination is rare and difficult to spot because the Higgs boson produced this way is often drowned out by a flood of other, more common particle collisions that look almost identical.

A team of researchers has developed a new method to separate this rare signal from the overwhelming background noise using advanced computer learning. They simulated millions of proton-proton collisions, similar to those happening at the Large Hadron Collider, to create a training ground for a digital brain. This brain, a deep neural network, was taught to recognize the subtle differences between the rare event where a Higgs boson and a top quark appear together and the common events that mimic them. The researchers found that by using a specialized technique to handle missing information within the data, their system could identify the rare events with much greater accuracy than previous methods, significantly improving the chances of observing this elusive process in real experiments.

The challenge in finding this specific particle interaction is that it happens very rarely. In the high-energy collisions of the Large Hadron Collider, the process where a Higgs boson is produced with a single top quark is suppressed by a quantum mechanical effect known as destructive interference. This means that the natural laws of physics conspire to make this event happen far less often than other similar processes. When it does occur, the particles it produces decay into a spray of other particles, including jets of energy and a charged lepton. However, many other common collisions produce almost the exact same spray of debris. It is like trying to find a single specific grain of sand on a beach that is constantly being reshaped by waves, where the grain you are looking for looks exactly like the millions of others around it.

To solve this, the researchers turned to a machine learning approach. Instead of trying to set rigid rules to cut away the unwanted events, they built a neural network, a type of computer program inspired by the human brain, capable of learning complex patterns. They fed the network data from millions of simulated collisions, labeling each one as either the rare signal they wanted or the common background noise they wanted to ignore. The network learned to weigh dozens of different measurements from each collision, such as the energy and direction of the particles, to decide whether an event was likely to be the rare signal.

A key innovation in this study was how the network handled events where certain particles were missing. In some collisions, the detectors might not catch every single particle, leaving gaps in the data. Traditional methods often struggle with these gaps, either discarding the event or treating the missing data as a zero, which can confuse the computer. The researchers designed a special mechanism that tells the network to effectively ignore the missing parts of the data without letting the computer learn from the empty space. This allowed the network to focus only on the particles that were actually present, making its decisions much sharper and more reliable.

The team tested their new method against a standard approach that relies on simple cuts, like ignoring any event that doesn't meet a specific energy threshold. The results showed a dramatic improvement. Using their advanced neural network, the researchers were able to isolate the rare signal with a statistical significance of 3.73σ using a standard counting method, and 4.64σ using a more sophisticated shape-based analysis. In the world of particle physics, this level of improvement is crucial because it moves the observation of this rare process from simple preselection to a level of strong evidence, though it still falls short of the 5σ threshold required for a definitive discovery.

The study was conducted using computer simulations based on an integrated amount of data equivalent to what the Large Hadron Collider is expected to collect over several years of operation. While these results are based on simulations and not yet on real detector data, they provide a strong roadmap for future experiments. The researchers noted that real-world conditions, such as the imperfect efficiency of the detectors in identifying certain particles, would likely reduce the final sensitivity, but the method they developed offers a powerful tool to maximize the chances of success. By refining how they analyze the data, scientists hope to soon confirm the properties of the Higgs boson in this rare channel, potentially revealing new insights into the fundamental structure of the universe.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →