The 3rd Quantum Computing Workshop

AI Optimization and Forecasting across Industries:

Digital and Quantum Computing

April 9-12, 2024

 I4-01-03 Seminar Room, NUS, Singapore

The 2024 workshop aims to serve as a dynamic platform for exchanging the latest developments and insights in AI optimization and forecasting. Our focus will be on accelerating advancements and growth across four pivotal industries: digital finance, energy, public health, and transport. We will explore recent advances in digital computing and delve into the potential of quantum computing as the next generation of computing technology. Our goal is to foster a deeper understanding of how AI optimization and forecasting can revolutionize various sectors through transformative science and research. This significant gathering of leading experts from academia and industry will focus on digital and quantum computing, with an emphasis on theory, methods, and algorithms.

Programme

• Day 1 (Tuesday, April 9, 2024)  Hands-on Tutorial on Quantum Computing and Algorithms
• [April 10 is PH (Hari Raya Puasa) in Singapore]
• Day 2 (Thursday, April 11, 2024): Company/Lab Tours and Frontiers in Tech Roundtable: Bridging Science and Industry [By invitation only]   
  
• Day 3 (Friday, April 12, 2024) Workshop Talks and Networking Sessions

Hands-on Tutorial: Introduction to Quantum Computing & Financial Optimization with Qiskit

Dr. Stefan Wörner: Principal Research Scientist and Manager of the Quantum Computational Science group of IBM Quantum at IBM Research Europe – Zurich.

Prerequisite: Python
Venue: I4.0 Seminar Room 01-03
Time: Tuesday, April 9 @ 2:00-5:00PM

In this hands-on tutorial, we cover the basics of quantum computing, its applications, e.g. in finance, and how to program quantum computers with Qiskit. We introduce qubits and quantum circuits, first quantum algorithms, and show how to solve illustrative optimization problems, such as portfolio optimization. To conclude, first quantum circuits will be executed on real quantum computers via the could and we discuss corresponding challenges.

Frontiers in Tech Roundtable: Bridging Science and Industry (By Invitation Only)

Roundtable Panelist (alphabetical order)

– Ronald Chen (Head of Business Solutions, Analytics Centre of Excellence, DBS)

– Dr. Lee Shiang Long (Group CTO, ST Engineering)

– Ruth Lim (Head of R&D, Toll Group)

– Dr. Sunil Sivadas (Distinguished Researcher, NCS)

– Adeline Tay (Senior Director, Smart Manufacturing and AI, Micron)

– Dr. Sy Bor Wang (Lead Quantitative Strategist, GIC)

– Dr. Stefan Wörner (Principal Research Scientist, IBM Research Europe – Zurich)

– Dr. Alexander Yip (Clinical Director Healthcare Design, Head of Gastroenterology & Hepatology, Alexandra Hospital NUHS)

Workshop: AI Optimization and Forecasting across Industries: Digital and Quantum Computing

Venue: I4.0 Seminar Room 01-03
Time: Friday, April 12

Speakers (alphabetical order)

– Ying Chen (National University of Singapore)
– Paul Robert Griffin (Singapore Management University)
– Xue Song Geng (Singapore Management University)
– Thorsten Koch (Technische Universität Berlin & Zuse Institute Berlin)
– Stefan Lessmann (Humboldt-Universität zu Berlin)
– Ariel Neufeld (Nanyang Technological University)
– Patrick Rebentrost (National University of Singapore)
– Stefan Wörner (IBM Research Europe – Zurich)

Talk Title: Leveraging Quantum Algorithms for Large-Scale Dynamic Portfolio Optimization with Market Frictions

Abstract: We investigate the numerical performance of quantum computing algorithms against traditional digital computing alternatives using real world data sets in solving dynamic trading strategies for large-scale portfolio allocation. This involves a multiperiod discrete portfolio optimization problem that takes into account various market frictions, ranging from transaction costs to path-dependent capital constraints. These frictions pose significant challenges in formulating and solving dynamic portfolio optimization problems. Our research leverages advanced mathematical tools and quantum computing technologies to formulate the problem within the framework of Quadratic Unconstrained Binary Optimization (QUBO) and solve it using quantum algorithms. Our findings indicate a marked improvement in efficiency and practical applicability, showcasing the potential of quantum mechanics to revolutionize financial decision-making processes in the presence of market frictions. This study not only contributes to the existing body of knowledge by providing a novel methodological framework but also opens new avenues for future research in the application of quantum computing in finance.

Talk Title: Quantum Machine Learning for Credit Scoring

Abstact: In this presentation we explore the use of quantum machine learning (QML) applied to credit scoring for small and medium size businesses (SMEs). A quantum/classical hybrid approach has been used with several models, activation functions, epochs, other parameters. Results are shown from the best model, using two quantum classifiers and a classical neural network, applied to data for companies in Singapore. We observe significantly more efficient training for the quantum models over the classical models for comparable prediction performance. Practical issues are also explored including a quadratic computational slow down with the number of qubits and a linear slow down with the number of blocks of classifiers using classical simulators. Running the models on real quantum computers is discussed including the number of times a circuit has to be executed. Surprisingly, a degradation in the accuracy was observed as the number of qubits was increased beyond 12 qubits and also with the addition of extra classifier blocks in the quantum model. Overall, we see great promise in this first in-depth exploration of the use of hybrid QML in credit scoring.

Talk Title: Business models with AI

In today’s rapidly evolving technological landscape, harnessing the commercial potential of AI requires a strategic understanding of effective business models. Successful integration of AI technologies and business hinges on a nuanced approach that aligns technology attributes with monetization logics. This entails a deeper understanding of how AI has transformed value creation process, value distribution among different parties on the value chain, cost structures, and revenue streams. The evolving nature of AI may demand novel business models. The talk will discuss this transformation and shed light on its implication on how firms can design business models to fully capitalize on the commercial possibilities presented by AI technologies.

Talk Title: Algorithmic Intelligence: A mostly discrete tour through challenges in AI Optimization

Abstract: We introduce Algorithmic Intelligence methods to tackle challenging real-world AI optimization problems in e.g. gas transport networks and terminals operations. We will start with an exploration of mixed-integer non-linear Optimization (MINLP) and mixed-integer linear Optimization (MILP),uncovering their pivotal roles in tackling complex industrial optimization challenges. Our tour will then take a brief interlude to examine linear programming (LP) and its relation to exact solutions needed in combinatorial auctions, chip design verification, and computational proofs. Afterward, we highlight recent algorithms, software, modeling, and parallel computing developments. Along the way, we will dive into the intricacies of Steiner tree problems and explore the fascinating field of Quadratic Unconstraint Binary Optimization (QUBO), including its potential relevance to quantum computing and its relevance to solving future challenges.

Talk Title: Transfer learning for credit risk modeling

Abstract: The paper provides the following contributions. First, we revisit credit scoring applications and identify use cases for transfer learning. We also elaborate on methodologies to transfer learn gradient boosting-based risk models. This illustrates how transfer learning can be put into practice using a learning algorithm with a proven track record in credit scoring. Next, we conduct a simulation study to systematically examine the effectiveness of transfer learning across relevant forms of data discrepancies between the source and target domains using synthetic data. Our analysis embraces various distribution shifts, including prior and covariate drift, concept drift, and data imbalances to offer crisp recommendations on when and when not to consider transfer learning. Last, we employ real-world data from installment loans and assess the merit of transfer learning corresponding scorecards by pretraining models on micro-lending data. Considering relevant settings of discrepancies in the feature sets, our experiments extend the simulation study and facilitate verifying that the advantages of transfer learning observed therein also translate into real-life applications.

Talk Title: Quantum Monte Carlo algorithm for solving Black-Scholes PDEs for high-dimensional option pricing in finance and its proof of overcoming the curse of dimensionality

Abstract: In this talk we present a quantum Monte Carlo algorithm which can solve high-dimensional Black-Scholes PDEs with correlation for high-dimensional option pricing. The payoff function of the option is of general form and is only required to be continuous and piece-wise affine (CPWA), which covers most of the relevant payoff functions used in finance. We provide a rigorous error analysis and complexity analysis of our algorithm. In particular, we prove that the computational complexity of our algorithm is bounded polynomially in the space dimension d of the PDE and the reciprocal of the prescribed accuracy ε and so demonstrate that our quantum Monte Carlo algorithm does not suffer from the curse of dimensionality.

Talk Title: Quantum linear algebra is all you need for Transformer architectures

Abstract: Generative machine learning methods such as large-language models are revolutionizing the creation of text and images. While these models are powerful they also harness a large amount of computational resources. The transformer is a key component in large language models that aims to generate a suitable completion of a given partial sequence. In this work, we investigate transformer architectures under the lens of fault-tolerant quantum computing. The input model is one where pre-trained weight matrices are given as block encodings to construct the query, key, and value matrices for the transformer. As a first step, we show how to prepare a block encoding of the self-attention matrix, with a row-wise application of the softmax function using the Hadamard product. In addition, we combine quantum subroutines to construct important building blocks in the transformer, the residual connection, layer normalization, and the feed-forward neural network. Our subroutines prepare an amplitude encoding of the transformer output, which can be measured to obtain a prediction. We discuss the potential and challenges for obtaining a quantum advantage.

Talk Title: Quantum Optimization in the Era of Quantum Utility

Abstract: During this talk, we first discuss the potential advantage of quantum computing in optimization from a complexity theoretic perspective and highlight the importance of quantum optimization heuristics. The ascent of quantum computers with 100+ qubits allows us to develop and test such heuristics in practice at a non-trivial scale. To this extent, we discuss Quantum Error Mitigation in the context of optimization and show recent results of a demonstration on 127 qubits.

_________________________________________________________________________________________________________

Organizer:

• Ying Chen

• Kian Guan Lim
• Patrick Rebentrost

_________________________________________________________________________________________________________

Coordinator:

• Hoa Nguyen
• Hongrui Zhang