| Lecture Note | Home > Lecture Note |
1. 1st week : ML, QC, QML
Myeonghun Park (Seoultech)
Title : Machine Learning basics
Abstract : We will study basic concepts of machine learning (ML) with toy examples (PyTorch library). I will discuss (1) kernel methods (2) deep neural network (3) convolutional neural network (4) recurrent neural network including some variations. I will also talk about the contributions from physics to the early stage of machine learning including classical simulated annealing algorithm for optimization and (restricted) Boltzmann machine etc. If time permits, I will briefly discuss graph neural networks.
* Code Link and Lecture Slide
Ramon Winterhalder (UC Louvain)
Title : Modern machine learning for particle physics
Abstract/Syllabus : In recent years, machine learning techniques have revolutionized the field of particle physics, enabling breakthroughs in data analysis, signal detection, and modeling complex systems. This lecture aims to provide an overview of three key topics in modern machine learning for particle physics applications: Bayesian neural networks, generative models, and anomaly detection.
1. Bayesian Neural Networks: Bayesian neural networks offer a probabilistic framework for training and inference in neural networks, allowing for uncertainty quantification and robust decision-making. We will explore the principles of Bayesian inference, discuss the advantages of Bayesian neural networks in handling limited data and training uncertainties, and demonstrate their application in tasks such as amplitude learning.
2. Generative Models I & II: Generative models play a vital role in understanding and simulating particle physics phenomena. In Part I, we will delve into the foundations of generative models, including variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows, and diffusion models. We will discuss their basic functionality and differences, and their ability to generate realistic images and model complex probability distributions. In Part II, we will focus on the application of generative models in tasks such as event generation and simulation-based inference, showcasing their potential to improve the precision of theoretical predictions and enhance experimental sensitivity.
3. Anomaly Detection: Anomaly detection plays a crucial role in particle physics experiments for identifying rare and elusive events that may indicate physics beyond the Standard Model. We will explore various approaches for anomaly detection, including unsupervised learning, density estimation techniques, and classification. Additionally, we will discuss the challenges of anomaly detection in high-energy physics experiments and showcase real-world examples of anomaly detection applied to large-scale particle physics datasets.
* Lecture 01, 02, 03, 04 and Slide
Sergei Gleyzer (University of Alabama)
Title : Introduction to quantum machine learning
Abstract : I will focus on quantum machine learning (e.g. variational circuits) and connection to classical ML algorithms. I’ll discuss different types of quantum machine learning (QML) algorithms in detail from QSVMs to QDNN, QCNN, quantum graphs, quantum generative models (QVAEs, QGANs), data re-uploading etc.
* Lecture (Tue) , Lecture (Thu)
Jack Y. Araz (Durham U)
Title : Introduction to quantum computation
Abstract : One can observe similarities between statistics and quantum mechanics in many ways. In this lecture series, we will discuss the gray area between Quantum Machine Learning and quantum many-body physics while trying to emphasise similarities between the two applications. We will start our lectures with Tensor Networks, a classical approach to constructing quantum many-body systems, and learn how to use them for data analysis problems. Furthermore, we will discuss the transition between Tensor Networks and Quantum Circuits while going through the limitations of each computation technique.
Tutorials for KIAS QUC Summer School on A.I. in High Energy Physics
https://github.com/jackaraz/kias_quc_qml/tree/main
· examples about tensor contractions.
https://colab.research.google.com/github/jackaraz/kias_quc_qml/blob/main/examples_tns.ipynb
· ML examples using tensor networks.
https://colab.research.google.com/github/jackaraz/kias_quc_qml/blob/main/tensor_contractions.ipynb
· examples on variational quantum eigensolver.
https://colab.research.google.com/github/jackaraz/kias_quc_qml/blob/main/time_evolution.ipynb
· example on variational quantum thermaliser.
https://colab.research.google.com/github/jackaraz/kias_quc_qml/blob/main/vqe.ipynb
· examples on quantum simulation.
https://colab.research.google.com/github/jackaraz/kias_quc_qml/blob/main/vqt.ipynb
* Lecture 1 Tensor Networks, Lecture 2 Quantum Computing
*Project Discussion*
https://docs.google.com/document/d/1_3D5qDIKQrxy6hnbSW63YFIJfcoJBgAFTOY3UWF7ljo/edit?usp=sharing
2. 2nd week
Eung Jin Chun (KIAS)
Title : Strong CP proglem and PQ mechanism
Fuminobu Takahashi (Tohoku U)
Title : Axions in astrophysics and cosmology
Edward (Rocky) W. Kolb (FNAL and U. of Chicago)
Title : Cosmological Gravitational particle productions
Yael Shadmi (Technion)
Title : On-shell tools for BSM physics
Reference :
General amplitude methods: Elvang & Huang, scattering amplitudes (bookor lecture notes, 1308.1697), Ch's 2, 3.
Massive amplitudes: Arkani-Hamed Huang & Huang, 1709.04891.
Massive amplitudes and EFTs: Shadmi & Weiss, 1809.09644; Durieux,Kitahara, Shadmi & Weiss 1909.10551
EFTs: Tim Cohen1903.03622
Seminar/Colloquium
Seth Lloyd (MIT) / Online
Title : Quantum computing past, present, and future
Abstract : The field of quantum computing is progressing rapidly, but to where? This talk reviews advances in quantum computing, in both hardware and algorithms, and discusses the technological challenges that need to be overcome to make large scale quantum computing a reality.
Tilman Plehn (Heidelberg) / Online
Title : LHC Physics as Data Science
Abstract : Modern machine learning is having significant impact on essentially all aspects of LHC physics. The simple reason is that LHC physics combines vast and highly complex data sets with precise first-principles predictions in a unique manner. I will introduce several ML-related aspects of LHC theory and show how we can benefit from new concepts and methods. This includes precision simulations including uncertainty estimates, inverted simulations and unfolding in a mathematically consistent manner, anomaly searches, and symbolic regression, to close the theory circle and return to formulas as the language of theoretical physics.
Donghui Jeong (Penn State Univ & KIAS)
Title : Parity violation in cosmology
Yannis K. Semertzidis (IBS-CAPP)
Title : The final frontier in axion dark matter search