Fraunhofer Institute for Industrial Mathematics ITWMQuantum Computing (QC) and Machine Learning (ML) are key technologies that will significantly change our technological landscape in the coming decades, and in some cases are already doing so today. In order to achieve competitive results in these fields, highly qualified experts with expertise in both areas are required. The module covers topics at the intersection of Quantum Computing and Machine Learning. It is aimed at people with a Quantum Computing background as well as people with a background in data science. Participants will gain the ability to successfully apply Machine Learning with Quantum Computers. To this end, numerous current methods are presented that enable them to react to future hardware advances and independently develop new QML algorithms. The concepts taught are illustrated with a large number of case studies from real applications and projects. A large part of the course is devoted to consolidating what has been learned with practical application examples.
Introduction to Quantum Machine Learning
Certified Data Scientist specialized in Quantum Machine Learning
Upon successful completion of this module, participants will have acquired the following learning outcomes:
Learning Objective:
The participants
- know the basic formal concepts of Quantum Computing (quantum state, bit vs. qubit, measurement)
- know the basic formal concepts of Machine Learning (objective function, model class, cross-validation, kernel function)
- learn to use ideas and building blocks of Quantum Computing for QML problems
Knowledge / Understanding
The participants
- can describe the quantum support vector machine method and use it in application cases
- understand the strengths, weaknesses and limitations of current QML methods
Abilities / Skills
The participants
- can read quantum circuits and create them independently
- are able to encode data on the Quantum Computer and subsequently analyze the encoding
- are able to apply hybrid quantum-classical optimization algorithms (e.g. Variational Quantum Eigensolver (VQE) and Quadratic Unconstrained Binary Optimization (QUBO))
- are able to create quantum clustering algorithms and implement them in practical examples
Projekt Partner
- Fraunhofer Institute for Industrial Mathematics ITWM(Coordinator und Content Creator)
- Fraunhofer Institut for Open Communication Systems FOKUS(Content Creator)
- Fraunhofer Institute for Intelligent Analysis and Information Systems IAIS (Content Creator)
- Fraunhofer Institute for Manufacturing Engineering and Automation IPA (Content Creator)
- Fraunhofer Institute for Applied Information Technology FIT
Target Group
- Experts from the fields of data science and machine learning
- Employees of technology companies, such as pharmaceutical and chemical companies
- Employees of government agencies interested in potential applications in the fields of cryptography and cyber security
- Employees of research institutions and students pursuing a master's degree or doctorate in fields such as computer science, physics, mathematics or data science who would also like to update their knowledge of QML
- Employees of research institutions and students who have previous experience in the field of Quantum Computing
Participants have the opportunity to obtain a personal certificate after completing the training by passing an examination. The certificate serves as a classification of the level of knowledge and as proof of the ability to deal with Quantum Computers in the field of Quantum Machine Learning. With the certificate, participants can present a standardized document for future jobs, which certifies the handling and especially the practical implementation of Quantum Computing.
Topics
- Part 1: Basics of Machine Learning/Data Science
- Data preprocessing
- Feature spaces
- Supervised learning, unsupervised learning
- Exemplary problems: classification, clustering
- Complexity
- Evaluation
- Part 2: Quantum Computing
- Basic theoretical concepts
- Different paradigms: Quantum Gate and Adiabatic
- Quantum Fourier transform
- Quadratic Unconstrained binary optimization (QUBO)
- Advantages over classical
Unit 2
- Clustering needed for Quantum Computing
- Grover algorithm
- Quantum k-Means
- SWAP test
Unit 3
- Parametrized quantum circuits
- Data encoding
- Analyzing parametrized quantum circuits
Unit 4
- Classical support vector machines and kernel trick
- Quantum feature maps
- Train quantum Kernels, kernel alignment
- Kernel based versus variational training in terms of circuit evaluations
Unit 5
- Neural networks
- Quantum neural networks (QNNs)
- Use cases of QNNs
- Potential quantum advantages of QNNs