Sequence Processing with Quantum Tensor Networks
For the first time, Quantinuum researchers have run scalable quantum natural language processing (QNLP) models, able to parse and process real-world data, on a quantum computer. In a recent paper, the researchers define machine learning models for the task of classifying sequences – which can be anything from sentences in natural language, like movie reviews, to bioinformatic strings, like DNA sequences. Classifying sequences of symbols – letters, words, or longer fragments of text – is an obviously useful computational task, and has led to some of the decade’s biggest changes; we now see this technology in use in everything from chatbots to legal cases.
Current classical models, which are based on neural networks, primarily look at the statistical distributions of where words are put with respect to each other – they don’t really consider the structure of language a priori (they could, but they don’t). In contrast, syntactic information scaffolds Quantinuum’s new quantum models, which are based on tensor networks, making them “syntax-aware”. Considering things like structure and syntax from the beginning allows scientists to create models with far fewer parameters, that require fewer gate operations to run, while allowing for interpretability thanks to the meaningful structure baked in from the start. Interpretability is the most pressing challenge in artificial intelligence (AI) — because if we don’t know why an algorithm has given an answer, we can’t trust it in critical applications, for instance in making medical decisions, or in scenarios where human lives are at stake.
Both neural and tensor networks can capture complex correlations in large data, but the way they do it is fundamentally different. In addition, since quantum theory inherently is described by tensor networks, using them to build quantum natural language processing models allows for the investigation of the potential that quantum processors can bring to natural language processing specifically, and artificial intelligence in general.
Thanks to best-in-class features like mid-circuit measurement and qubit reuse on Quantinuum’s H2-1 quantum processor, they were able to fit much larger circuits than one might naively expect. For example, the researchers were able to run a circuit that would normally take 64 qubits on only 11 qubits. Combined with the reduced number of gates required, these models are entirely feasible on current quantum hardware.
This paper shows us that we can run, train, and deploy QNLP models on present-day quantum computers. When compared to neural-network-based classifiers, the quantum model does just as well on this task in terms of prediction accuracy. What’s more, this work encourages the exploration of quantum language models, as sampling from quantum circuits of the types used in this work could require polynomially fewer resources than simulating them classically.
About Quantinuum
Quantinuum, the world’s largest integrated quantum company, pioneers powerful quantum computers and advanced software solutions. Quantinuum’s technology drives breakthroughs in materials discovery, cybersecurity, and next-gen quantum AI. With over 500 employees, including 370+ scientists and engineers, Quantinuum leads the quantum computing revolution across continents.
For a novel technology to be successful, it must prove that it is both useful and works as described.
Checking that our computers “work as described” is called benchmarking and verification by the experts. We are proud to be leaders in this field, with the most benchmarked quantum processors in the world. We also work with National Laboratories in various countries to develop new benchmarking techniques and standards. Additionally, we have our own team of experts leading the field in benchmarking and verification.
Currently, a lot of verification (i.e. checking that you got the right answer) is done by classical computers – most quantum processors can still be simulated by a classical computer. As we move towards quantum processors that are hard (or impossible) to simulate, this introduces a problem: how can we keep checking that our technology is working correctly without simulating it?
We recently partnered with the UK’s Quantum Software Lab to develop a novel and scalable verification and benchmarking protocol that will help us as we make the transition to quantum processors that cannot be simulated.
This new protocol does not require classical simulation, or the transfer of a qubit between two parties. The team’s “on-chip” verification protocol eliminates the need for a physically separated verifier and makes no assumptions about the processor’s noise. To top it all off, this new protocol is qubit-efficient.
The team’s protocol is application-agnostic, benefiting all users. Further, the protocol is optimized to our QCCD hardware, meaning that we have a path towards verified quantum advantage – as we compute more things that cannot be classically simulated, we will be able to check that what we are doing is right.
Running the protocol on Quantinuum System Model H1, the team ended up performing the largest verified Measurement Based Quantum Computing (MBQC) circuit to date. This was enabled by our System Model H1’s low cross-talk gate zones, mid-circuit measurement and reset, and long coherence times. By performing the largest verified MBQC computation to date, and by verifying computations significantly larger than any others to be verified before, we reaffirm the Quantinuum Systems as best-in-class.
Particle accelerators like the LHC take serious computing power. Often on the bleeding-edge of computing technology, accelerator projects sometimes even drive innovations in computing. In fact, while there is some controversy over exactly where the world wide web was created, it is often attributed to Tim Berners-Lee at CERN, who developed it to meet the demand for automated information-sharing between scientists in universities and institutes around the world.
With annual data generated by accelerators in excess of exabytes (a billion gigabytes), tens of millions of lines of code written to support the experiments, and incredibly demanding hardware requirements, it’s no surprise that the High Energy Physics community is interested in quantum computing, which offers real solutions to some of their hardest problems. Furthermore, the HEP community is well-positioned to support the early stages of technological development: with budgets in the 10s of billions per year and tens of thousands of scientists and engineers working on accelerator and computational physics, this is a ripe industry for quantum computing to tap.
As the authors of this paper stated: “[Quantum Computing] encompasses several defining characteristics that are of particular interest to experimental HEP: the potential for quantum speed-up in processing time, sensitivity to sources of correlations in data, and increased expressivity of quantum systems... Experiments running on high-luminosity accelerators need faster algorithms; identification and reconstruction algorithms need to capture correlations in signals; simulation and inference tools need to express and calculate functions that are classically intractable”
The authors go on to state: “Within the existing data reconstruction and analysis paradigm, access to algorithms that exhibit quantum speed-ups would revolutionize the simulation of large-scale quantum systems and the processing of data from complex experimental set-ups. This would enable a new generation of precision measurements to probe deeper into the nature of the universe. Existing measurements may contain the signatures of underlying quantum correlations or other sources of new physics that are inaccessible to classical analysis techniques. Quantum algorithms that leverage these properties could potentially extract more information from a given dataset than classical algorithms.”
Our scientists have been working with a team at DESY, one of the world’s leading accelerator centers, to bring the power of quantum computing to particle physics. DESY, short for Deutsches Elektronen-Synchrotron, is a national research center for fundamental science located in Hamburg and Zeuthen, where the Center for Quantum Technologies and Applications (CQTA) is based. DESY operates, develops, and constructs particle accelerators used to investigate the structure, dynamics and function of matter, and conducts a broad spectrum of interdisciplinary scientific research. DESY employs about 3,000 staff members from more than 60 nations, and is part of the worldwide computer network to store and analyze the enormous flood of data that is produced by the LHC in Geneva.
In a recent paper, our scientists collaborated with scientists from DESY, the Leiden Institute of Advanced Computer Science (LIACS), and Northeastern University to explore using a generative quantum machine learning model, called a “quantum Boltzmann machine” to untangle data from CERN’s LHC.
The goal was to learn probability distributions relevant to high energy physics better than the corresponding classical models. The data specifically contains “particle jet events”, which describe how colliders collect data about the subatomic particles generated during the experiments.
In some cases the quantum Boltzmann machine was indeed better, compared to a classical Boltzmann machine. The team is analyzed when and why this happens, understanding better how to apply these new quantum tools in this research setting. The team also studied the effect of the data encoding into a quantum state, noting that it can have a decisive effect on the training performance. Especially enticing is that the quantum Boltzmann machine is efficiently trainable, which our scientists showed in a recent paper published in Nature Communications Physics.
Find the Quantinuum team at this year’s SC24 conference from November 17th – 22nd in Atlanta, Georgia. Meet our team at Booth #4351 to discover how Quantinuum is bridging the gap between quantum computing and high-performance compute with leading industry partners.
The Quantinuum team will be participating various events, panels and poster sessions to showcase our quantum computing technologies. Join us at the below sessions:
Monday, Nov 18, 8:00 - 8:25pm, EST
Panel: KAUST booth 1031
Nash Palaniswamy, Quantinuum’s CCO, will join a panel discussion with quantum vendors and KAUST partners to discuss advancements in quantum technology.
Monday, Nov 18, 9:00 - 11:59pm, EST
Beowulf Bash: World of Coca-Cola
This year, we are proudly sponsoring the Beowulf Bash, a unique event organized to bring the HPC community together for a night of unique entertainment! Join us at the event on Monday, November 18th, 9:00pm at the World of Coca-Cola.
Wednesday, Nov 20, 3:30 – 5:00pm, EST
Panel: Educating for a Hybrid Future: Bridging the Gap between High-Performance and Quantum Computing
Vincent Anandraj, Quantinuum’s Director of Global Ecosystem and Strategic Alliances, will moderate this panel which brings together experts from leading supercomputing centers and the quantum computing industry, including PSC, Leibniz Supercomputing Centre, IQM Quantum Computers, NVIDIA, and National Research Foundation.
Thursday, Nov 21, 11:00 – 11:30am, EST
Presentation: Realizing Quantum Kernel Models at Scale with Matrix Product State Simulation
Pablo Andres-Martinez, Research Scientist at Quantinuum, will present research done in collaboration with HSBC, where the team applied quantum methods to fraud detection.