Searching for exotic particles in high-energy physics with deep learning
By Pierre Baldi, Peter Sadowski, Daniel Whiteson. 2014
1,320 citations recorded by the source corpus.
Collisions at high-energy particle colliders are a traditionally fruitful source of exotic particle discoveries. Finding these rare particles requires solving difficult signal-versus-background classification problems, hence machine-learning approaches are often used. Standard approaches have relied on ‘shallow’ machine-learning models that have a limited capacity to learn complex nonlinear functions of the inputs, and rely on a painstaking search through manually constructed nonlinear features. Progress on this problem has slowed, as a variety of techniques have shown equivalent performance. Recent advances in the field of deep learning make it possible to learn more complex functions and better discriminate between signal and background classes. Here, using benchmark data sets, we show that deep-learning methods need no manually constructed inputs and yet improve the classification metric by as much as 8% over the best current approaches. This demonstrates that deep-learning approaches can improve the power of collider searches for exotic particles. High-energy particle colliders are important for finding new particles, but huge volumes of data must be searched through to locate them. Here, the authors show the use of deep-learning methods on benchmark data sets as an approach to improving such new particle searches.
Subfields: Artificial Intelligence, Nuclear and High Energy Physics, AI for Physical Sciences
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Yoshua Bengio
- Quoc V. Le
- Percy Liang
- Mohit Bansal
- Sergey Levine
- Jianfeng Gao
- Ruslan Salakhutdinov
- Kyunghyun Cho
In Nuclear and High Energy Physics
- E. Metodiev
- Patrick T. Komiske
- Kyle Cranmer
- Pierre Baldi
- Peter Sadowski
- Tom Goldstein
- Gilles Louppe
- Siyuan Ma
In AI for Physical Sciences
- Sifan Wang
- Pierre Baldi
- Peter Sadowski
- Gilles Louppe
- Tom Goldstein
- Paris Perdikaris
- Wojciech Samek
- Kyle Cranmer
Citations in the corpus
- Decentralized gradient methods: does topology matter? (2020, neutral): “(10)Minimization of hinge loss function (with L2 regularization) for classification on the dataset SUSY from (uci) (S = 5 × 10(5), n = 18 (Baldi et al., 2014)).”
- Protection Against Reconstruction and Its Applications in Private Federated Learning (2018, background): “…hardware and access to substantial amounts of data has made fitting accurate models for image classification, text translation, physical particle prediction, astronomical observation, and other predictive tasks possible with accuracy that was previously completely infeasible [11, 7, 49].”
- BOHB: Robust and Efficient Hyperparameter Optimization at Scale (2018, background): “…units per layer) of a feed forward neural network for six different datasets gathered from OpenML (Van-schoren et al., 2014): Adult (Kohavi, 1996), Higgs (Baldi et al., 2014), Letter (Frey & Slate, 1991), MNIST (LeCun et al., 2001), Optdigits (Lichman, 2013), and Poker (Cattral et al., 2002).”
- Generative Counterfactual Introspection for Explainable Deep Learning (2019, background): “The recent success of deep neural networks has lead to many breakthroughs in various application domains [1–3].”
- Explaining nonlinear classification decisions with deep Taylor decomposition (2015, background): “research fields such as natural language processing [13, 14], human action recognition [15, 16, 17], or physics [18, 19].”
- Understanding and mitigating gradient pathologies in physics-informed neural networks (2020, background): “Keywords Deep learning · Differential equations · Optimization · Stiff dynamics · Computational physics 1 Introduction Thanks to breakthrough results across a diverse range of scientific disciplines [1, 2, 3, 4, 5], deep learning is currently influencing the way we process data, recognize patterns, ”
- Fast Two-Sample Testing with Analytic Representations of Probability Measures (2015, background): “Remarkably, such a loss of power is not the rule: for instance, when distinguishing signatures of the Higgs boson from background noise [2] (’Higgs dataset’), we observe that a test based on differences in smoothed empirical characteristic functions outperforms the quadratic-time MMD.”
- DeltaGrad: Rapid retraining of machine learning models (2020, methodology): “For any given delete/add rate, DeltaGrad achieves signifcant speed-ups (up to 2.6x for MNIST, 2x for covtype, 1.6x for HIGGS, 6.5x for RCV1) compared to BaseL.”
- On the Expressive Power of Deep Neural Networks (2016, background): “ST and CIFAR-10. 1 Introduction Neural network architectures have proven “unreasonably effective” [1, 2] on many tasks, including image classification [3], identifying particles in high energy physics [4], playing Go [5], and modeling human student learning [6]. Despite their power, we have limited u”
- DeepTox: Toxicity Prediction using Deep Learning (2016, background): “Similar to the successes in other fields (Dahl et al., 2012; Krizhevsky et al., 2012; Deng et al., 2013; Graves et al., 2013; Socher and Manning, 2013; Baldi et al., 2014; Sutskever et al., 2014), Deep Learning has increased the predictive performance of computational methods in toxicology.”
- Efficient Distributed SGD with Variance Reduction (2015, methodology): “Figure 3 shows results of our distributed experiments on the large datasets SUSY and MILLIONSONG.”
- Asynchronous Decentralized Online Learning (2021, background): “(i) The higgs dataset is a benchmark dataset in high-energy physics [4] for binary classification, which consists of 11 million instances with 28 features.”
- Learning Deep Kernels for Non-Parametric Two-Sample Tests (2020, neutral): “Higgs dataset (Baldi et al., 2014).”
- Query-Efficient Hard-label Black-box Attack: An Optimization-based Approach (2018, methodology): “In this experiment, we use two standard datasets: HIGGS [23] for binary classification and MNIST [19] for multi-class classification.”
- Jet Substructure Classification in High-Energy Physics with Deep Neural Networks (2016, background): “[16] Pierre Baldi, Peter Sadowski, and Daniel Whiteson.”
- On Embeddings for Numerical Features in Tabular Deep Learning (2022, neutral): “In this section, we empirically evaluate the techniques discussed in section 3 and compare them with Gradient Boosted Decision Trees to check the status quo of the “DL vs GBDT” competition.”