PyTorchKR Papers

최신 논문

총 47편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1509.02216

Fuzzy Jets

Collimated streams of particles produced in high energy physics experiments are organized using clustering algorithms to form jets. To construct jets, the experimental collaborations based at the Large Hadron Collider (LHC) primarily use agglomerative hierarchical clustering schemes known as sequential recombination. We propose a new class of algorithms for clustering jets that use infrared and collinear safe mixture models. These new algorithms, known as fuzzy jets, are clustered using maximum likelihood techniques and can dynamically determine various properties of jets like their size. We show that the fuzzy jet size adds additional information to conventional jet tagging variables. Furthermore, we study the impact of pileup and show that with some slight modifications to the algorithm, fuzzy jets can be stable up to high pileup interaction multiplicities.

arXiv:1511.05190

Jet-Images -- Deep Learning Edition

Building on the notion of a particle physics detector as a camera and the collimated streams of high energy particles, or jets, it measures as an image, we investigate the potential of machine learning techniques based on deep learning architectures to identify highly boosted W bosons. Modern deep learning algorithms trained on jet images can out-perform standard physically-motivated feature driven approaches to jet tagging. We develop techniques for visualizing how these features are learned by the network and what additional information is used to improve performance. This interplay between physically-motivated feature driven tools and supervised learning algorithms is general and can be used to significantly increase the sensitivity to discover new particles and new forces, and gain a deeper understanding of the physics within jets.

arXiv:1703.03507

Decorrelated Jet Substructure Tagging using Adversarial Neural Networks

We describe a strategy for constructing a neural network jet substructure tagger which powerfully discriminates boosted decay signals while remaining largely uncorrelated with the jet mass. This reduces the impact of systematic uncertainties in background modeling while enhancing signal purity, resulting in improved discovery significance relative to existing taggers. The network is trained using an adversarial strategy, resulting in a tagger that learns to balance classification accuracy with decorrelation. As a benchmark scenario, we consider the case where large-radius jets originating from a boosted resonance decay are discriminated from a background of nonresonant quark and gluon jets. We show that in the presence of systematic uncertainties on the background rate, our adversarially-trained, decorrelated tagger considerably outperforms a conventionally trained neural network, despite having a slightly worse signal-background separation power. We generalize the adversarial training technique to include a parametric dependence on the signal hypothesis, training a single network that provides optimized, interpolatable decorrelated jet tagging across a continuous range of hypothetical resonance masses, after training on discrete choices of the signal mass.

arXiv:1711.08811

Machine Learning Algorithms for $b$-Jet Tagging at the ATLAS Experiment

The separation of $b$-quark initiated jets from those coming from lighter quark flavors ($b$-tagging) is a fundamental tool for the ATLAS physics program at the CERN Large Hadron Collider. The most powerful $b$-tagging algorithms combine information from low-level taggers, exploiting reconstructed track and vertex information, into machine learning classifiers. The potential of modern deep learning techniques is explored using simulated events, and compared to that achievable from more traditional classifiers such as boosted decision trees.

arXiv:1806.01263

Infrared Safety of a Neural-Net Top Tagging Algorithm

Neural network-based algorithms provide a promising approach to jet classification problems, such as boosted top jet tagging. To date, NN-based top taggers demonstrated excellent performance in Monte Carlo studies. In this paper, we construct a top-jet tagger based on a Convolutional Neural Network (CNN), and apply it to parton-level boosted top samples, with and without an additional gluon in the final state. We show that the jet observable defined by the CNN obeys the canonical definition of infrared safety: it is unaffected by the presence of the extra gluon, as long as it is soft or collinear with one of the quarks. Our results indicate that the CNN tagger is robust with respect to possible mis-modeling of soft and collinear final-state radiation by Monte Carlo generators.

arXiv:2606.21790v1cs.LG

로렌츠 등변 제트 태거는 무엇을 학습하는가?

What Do Lorentz-Equivariant Jet Taggers Learn?

로렌츠 등변 제트 태거가 내부적으로 무엇을 학습하는지, L-GATr, L-GATr-slim, LLoCa-T를 포함한 다섯 개 모델에 대해 등변성 테스트, 선형 프로브(linear probe), grade 소거를 통해 조사하였다. 선형 프로브는 등변 모델이 기준계에 의존하는 의사빠르기(pseudorapidity)를 0으로 억압하면서 제트 질량과 N-subjettiness를 강하게 인코딩함을 보여준다. L-GATr에 대한 grade 소거는 top 쿼크 태깅에서 이벌렉터(bivector) 채널은 무시할 만한 반면, 벡터형 채널은 지배적이지만 시드(seed)에 따라 달라짐을 밝히며, 이는 네트워크가 여러 표현 경로를 활용하고 있음과 일치한다. 이러한 결과는 등변 태거에서 어떤 물리적 특징과 대수적 grade 구조가 판별 정보를 담는지 특성화하며, 향후 이러한 모델의 개발에 기여할 수 있다.

arXiv:2606.17500v1cs.LG

재구성 가능 컴퓨팅 챌린지: Versal AI 엔진에서의 제트 태깅 트랜스포머

Reconfigurable Computing Challenge: Transformer for Jet Tagging on Versal AI Engines

트랜스포머 기반 모델은 CERN LHC에서 제트 태깅을 위해 강력한 성능을 발휘하지만, 저지연이 요구되고 자원이 제약된 트리거 시스템에 이를 배포하는 것은 어렵습니다. 우리는 AMD Versal AI 엔진(AIE)에서 제트 태깅을 위한 양자화된 정수 전용 트랜스포머의 초기 구현을 제시하며, 밀집(dense) 및 멀티 헤드 어텐션(MHA) 레이어를 AIE 타일에 매핑했습니다. 주요 기여는 트랜스포머 레이어를 구성 가능한 AIE 빌딩 블록으로 표현하고, 고수준 Python 모델 설명으로부터 해당 Vitis 그래프 코드를 자동으로 생성하는 재사용 가능한 소프트웨어 프레임워크입니다. 이 프레임워크는 향후 연구를 위한 기반을 제공하며, https://github.com/KastnerRG/particle_transformer_aie에서 오픈 소스 소프트웨어로 공개되었습니다.