Electronics Engineering | Course catalogue

QUANTUM COMPUTING AND NEURAL NETWORKS

Course objectives

KNOWLEDGE AND UNDERSTANDING The student will acquire knowledge of the basic notions regarding the design and implementation of quantum algorithms and quantum computing architectures for machine learning and artificial intelligence, in order to deal with variational quantum circuits and quantum neural networks learning. This will be based on the study of computational models, circuits and architectures along their universality, as well as on the explanation of the main algorithmic techniques exploiting quantum physics using model abstraction, in order to solve hard computational problems. The fundamentals of data-driven learning approaches will be acquired for applications to real-world problems, with specific implementations using quantum circuits and quantum neural networks along with the use of existing software platforms. CAPABILITY TO APPLY KNOWLEDGE AND UNDERSTANDING Solution of problems related to the design, implementation and testing of quantum computing architectures and quantum machine learning computational models for the solution of both supervised and unsupervised learning problems, such as optimization, prediction, clustering and classification, in real-world applications concerning signal, data and information processing. The main objective is to provide the student with the ability to understand and achieve quantum advantage in applications related to data-driven learning problems such as time series analysis, Hyperdimensional Computing, and eXplainable AI, considering several real domains pertaining to energy, aerospace, earth observation, behavioral analysis, bioengineering, finance, fraud detection, and so forth. MAKING AUTONOMOUS JUDGEMENTS Through a systematic laboratory activity, during which the methodologies related to the design and implementation of quantum computing architectures and quantum machine learning models such as quantum neural networks will be considered, the student will integrate the acquired knowledge to manage the complexity of inductive learning mechanisms and the actual limits imposed by currently adopted Noisy Intermediate-Scale Quantum (NISQ) devices, even starting from the limited information due to the practical organization of the course. COMMUNICATE SKILLS Quantum technologies and quantum information processing algorithms are rapidly evolving, considering the actual scenario based on near-term devices and hybrid quantum-classical approaches. Following this course, the student will be able to communicate the knowledge acquired to specialist and non-specialist interlocutors in the fields of research and work in which she/he will carry on the subsequent scientific and/or professional activities, also considering technological and sustainability issues. LEARNING SKILLS The adopted teaching methodology requires an autonomous and self-managed study activity during the development of monothematic homework for didactic and/or experimental investigation, i.e., in a vertical way on some specific theoretical and applicative topics using, for instance, available cloud-based quantum systems like IBM’s Quantum Experience Platform, as well as quantum simulators like Qiskit, Pennylane and Flax.

Channel 1

MASSIMO PANELLA Lecturers' profile

Program - Frequency - Exams

Course program

Introduction to Machine Learning. Artificial intelligence, machine learning and neural networks. Basic concepts of supervised and unsupervised learning. Overview of applications of machine learning and deep learning. Applications of machine learning to electrical and electronic systems. Practical approaches to the use of neural networks. Perceptrons. Data-driven learning, gradient descent and backpropagation. Model validation, overfitting and underfitting. Regularization. Data preparation. Introduction to time series prediction. Shallow neural networks. Radial Basis Function (RBF), Fuzzy Inference System (FIS) and neurofuzzy networks ANFIS, Extreme Learning Machine (ELM) and Random Vector Functional-Link (RVFL), Echo State Network (ESN). Basic concepts of randomization. Deep neural networks. Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Convolutional Neural Network (CNN). Computational aspects and hardware implementation of neural networks. Introduction to hyperdimensional computing. Circuit architectures for machine learning. Vector processor design methods. Dependence Graph (DG) and dependency analysis. Projection and scheduling, canonical mapping from DG to SFG, examples. Systolic arrays, fused multiply-add (FMA), pipelining, systolization. Parallel calculation on multi-core architectures, FPGA, GPU, TPU. Machine learning applied to computational architectures, circuits and systems. Main concepts, case studies, problems, solutions and future developments. Machine learning applied to electrical and environmental systems. Intelligent sensors, fault diagnosis, battery design and control systems. Machine learning applied to energy networks. Prediction from renewable sources, intelligent energy infrastructures, smart grids. Introduction to quantum computing. Quantum gates and arrays. Quantum algorithms for optimization and information processing (QFFT, Grover, Schor). Quantum machine learning. Quantum neural networks. Overview of the implementation on optical and optoelectronic devices. Practical exercises (hands-on) using Matlab: deep learning; GPU and parallel computing; prediction of energy time series; quantum computing and quantum machine learning. Applications and case studies: energy prediction from renewable sources, intelligent energy systems, Smart Grids; applications to real data (logistics, economical, biomedical, mechatronics, environmental and aerospace, etc.); biometrics and behavioral analysis; implementation of neural and neurofuzzy networks on microcontrollers, embedded systems and smart sensors.

Prerequisites

Basic knowledge of electronic and signal processing systems.

Books

H.T. Kung, R. Sproull, G. Steele, VLSI Systems and Computations, Springer. I. Goodfellow, Y. Bengio, A. Courville, Deep Learning, MIT Press. Teaching material provided through the course's web site.

Frequency

Compulsory attendance is not required.

Exam mode

Oral questions on the topics of the course. Didactic/experimental homeworks on a course topic.

Bibliography

H.T. Kung, R. Sproull, G. Steele, VLSI Systems and Computations, Springer. I. Goodfellow, Y. Bengio, A. Courville, Deep Learning, MIT Press S.O. Haykin, Neural Networks and Learning Machines (3rd Ed.), Pearson C.M. Bishop, Pattern Recognition and Machine Learning, Springer T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning (2nd Ed.), Springer Series in Statistics

Lesson mode

Teaching in presence. In cases of force majeure (health emergencies, etc.) the remote and/or blended methods will also be activated as per current rules.