Key Information:
What do we mean by quantum computer vision?
The aforementioned premises of quantum computing and annealing has led to the popularisation of quantum computer vision (QCV), where researchers started to port existing computer vision problems into forms amenable to quantum computation. However, reframing the existing problems in the context of this new computing paradigm is not trivial. For instance, existing literature relaxes most of the (discrete) combinatorial search problems to continuous ones, whereas quantum annealing is great at optimisation on discrete binary variables. Hence, oftentimes we are required to revise the problem formulations at hand altogether. The silver lining of this seemingly bad news is the novel research directions it opens up. In the past few years, we have started to witness the development of the first archetypes of this new area. Modern quantum computation has penetrated the realm of computer vision and machine learning in various subjects of study from point cloud registration to multi-object tracking with well supported industry grade development platforms such as D-Wave Ocean, Tensorflow Quantum, IBM Qiskit and Amazon Braket.
Why organize QCVML now?
We are optimistic that the quantum revolution is around the corner and the time has come. However, are we, as a community, prepared for this disruption? Our proposed workshop, QCVML (Quantum Computer Vision and Machine Learning) will be dedicated to investigating computer vision and machine learning problems, theoretically and experimentally, through the lens of practical quantum computation. Our main objective is to gather together industry experts, academic researchers, and CV-practitioners of quantum computing into a lively environment for discussing methodologies and challenges raised in exploiting these new types of computing devices; as a targeted topic venue, this workshop will offer participants a unique opportunity to network with a diverse but focused research community.
What is a quantum computer?
A quantum computer (QC) is a computing machine which takes advantage of quantum effects such as quantum superposition, entanglement, tunnelling and contextuality to solve problems notoriously difficult (belonging to challenging complexity classes such as NP) for a classical computer. Thanks to the exponentially increasing investment in the technology, quantum computers are gradually moving from the realms of theory towards actual devices. Albeit restricted, experimental realisations of numerous quantum algorithms have demonstrated improved computational performance finally reaching the desired supremacy in recent years. To date, IBM has constructed a device with 65 qubits and D-Wave has constructed a computer based on quantum annealing having >5000 qubits. Quantum computers offer speed-ups on certain specific algorithms including well known examples such as Shor’s algorithm, the quantum fourier transform (QFT), and, in the case of quantum annealing, solving non-convex, quadratic unconstrained binary optimization (QUBO) problems. This makes them amenable for deployment in the context of machine learning and computer vision, which demand significant amounts of compute resources. For example, visual perception algorithms need to process millions of pixels, or modern deep learning frameworks have millions of parameters to optimise for. The application of quantum computation to computer vision and machine learning is very interesting and intriguing because of both speed-up benefits and global optimality guarantees it offers.What do we mean by quantum computer vision?
The aforementioned premises of quantum computing and annealing has led to the popularisation of quantum computer vision (QCV), where researchers started to port existing computer vision problems into forms amenable to quantum computation. However, reframing the existing problems in the context of this new computing paradigm is not trivial. For instance, existing literature relaxes most of the (discrete) combinatorial search problems to continuous ones, whereas quantum annealing is great at optimisation on discrete binary variables. Hence, oftentimes we are required to revise the problem formulations at hand altogether. The silver lining of this seemingly bad news is the novel research directions it opens up. In the past few years, we have started to witness the development of the first archetypes of this new area. Modern quantum computation has penetrated the realm of computer vision and machine learning in various subjects of study from point cloud registration to multi-object tracking with well supported industry grade development platforms such as D-Wave Ocean, Tensorflow Quantum, IBM Qiskit and Amazon Braket.Why organize QCVML now?
We are optimistic that the quantum revolution is around the corner and the time has come. However, are we, as a community, prepared for this disruption? Our proposed workshop, QCVML (Quantum Computer Vision and Machine Learning) will be dedicated to investigating computer vision and machine learning problems, theoretically and experimentally, through the lens of practical quantum computation. Our main objective is to gather together industry experts, academic researchers, and CV-practitioners of quantum computing into a lively environment for discussing methodologies and challenges raised in exploiting these new types of computing devices; as a targeted topic venue, this workshop will offer participants a unique opportunity to network with a diverse but focused research community.Our invited speakers come from top research institutions and companies around the globe, and are leading figures in the topics covered by the workshop. This diverse selection will prove valuable for academic as well as industry researchers and practitioners. Both practical and theoretical aspects of quantum computing and its use in computer vision will be covered by the invited lecturers. We will make a selection out of the confirmed speakers, with a potential to include unconfirmed speakers. Both of these are listed below.
We have a packed and exciting day ahead of us!
We gladly welcome you to our workshop, QCVML 2024.
Building Robust Quantum ML System Software Stack
The field of quantum computing has enjoyed extraordinary advances in the last two decades, including the physical implementation and experimental demonstration of medium-scale quantum computers. While these advances continue to be celebrated widely, computational scientists and ML engineering continue to struggle to make meaningful use of existing quantum computers - aka Noisy Intermediate-Scale Quantum (NISQ) machines. This is because NISQ devices are highly error-prone and produce output that can be far from the correct output of the quantum algorithms. In this talk, I will primarily discuss the role of quantum ML system software in making erroneous quantum devices more usable and meaningful. If you have not written a quantum program yet and you love doing AI, I’ll show you that there is a whole exciting universe of quantum computers waiting for you!
Quantum Simulation of Anisotropic Image Diffusion
Simulating quantum mechanical systems on classical computers is notoriously difficult due to the exponential scaling of their state spaces. Quantum computers, however, can naturally exploit this exponential complexity to efficiently simulate quantum systems. While quantum simulations involve solving the Schrödinger equation, this approach can be generalised to solve other partial differential equations (PDEs) with real world applications, including financial options pricing and image processing. In this presentation, we will briefly introduce gate-based quantum computing and provide an overview of quantum simulation algorithms. We then explain our new methodology for solving PDEs using quantum simulation providing examples relevant to heat transfer and financial options trading. Finally, we demonstrate an example of PDE based image processing by simulating the anisotropic diffusion equation using our methodology.
We are proudly serving Italian coffee to accompany fruitful discussions.
We invite you to join us for a brief overview of recently published papers on quantum computer vision.
Quantum Binary Optimization
Binary optimization is an omnipresent problem in computer vision. It helps, for example, to model decision problems such as labeling, matching, tracking, clustering, and more. However, solving binary optimization problems classically is challenging due to their discrete and combinatorial nature. Over the last few years, quantum computing, especially adiabatic quantum computing, has shown promising results in solving binary problems, raising the question of whether more efficient solvers exploiting quantum properties could be designed. In this talk, I discuss quantum solvers for binary optimization problems. Specifically, I will present a variational solver that splits the task hybridly into quantum and classical parts, where a quantum computer executes only some specific, ideally quantum-native tasks, and a classical computer runs an optimization procedure to optimize the objective function. I will show how to evaluate the objective and even compute its analytical gradient on the quantum hardware, allowing the use of gradient-based sub-solvers for optimizing the objective.
Advancing Diffusion Models through Quantum Computing and Flow Matching
Untill next time!
QCVML 2024 hosts the submission of posters for works targeting the computer vision community. Researchers with recent publications or ongoing projects with early results in the following areas are encouraged to submit a poster summarizing their work:
Posters proposals should use the QCVML template and be submitted to qcvml.workshop@gmail.com. The deadline for submission is 15.09.2024 23:59 AOE. Posters following the topic and presentation guidelines will be invited to be presented during our poster session to spark discussions.
Besides this, we will update and maintain a curated collection of recent papers that target the computer vision community and are relevant to the field and invite previously published and peer-reviewed works to be submitted. The following papers have previously been published and peer-revieweed at major computer vision and machine learning conferences.
Paper Collection QCVML Poster TemplateHere are the diligent people behind QCVML 2024.
A small community in Quantum Computer Vision has formed over the last years and is steadily growing. With quantum computers being easily accessible by now, we believe that the time is right to show the potential of quantum computer vision and to bring together researchers from both fields.
Attending the workshop is a great opportunity to gain a glimpse on the basics in quantum computing. Furthermore, invited talks from academia and industry will give an overview of the current developments of the field and our poster session will give you the opportunity to see the current state-of-the-art collected in one place.
Yes, QCVML will be held fully in-person in September at ECCV 2024. While there will be no live streaming, we will record the talks and make them available on the website. Furthermore, an extended version of the poster session will be presented on the website.
Follow the recent hapennings here.