Q2B26 Tokyo | Wesley Dyk, Quantum Computing, Inc. and Paul Griffin, Singapore Management University

Value of the Information & Strength of the Argument

The value of the information is significant for those interested in practical applications of quantum computing. The presentation provides concrete results from a real-world financial problem, showing a clear improvement in fraud detection metrics. The argumentation is solid, with a clear explanation of the problem, the methodology, and the results. However, the presentation is somewhat promotional, as it is given by a company representative. The speakers do not delve deeply into the theoretical reasons for the quantum advantage, but they do offer a plausible explanation: the quantum optimizer finds a set of good solutions rather than a single optimal one, leading to more robust models. The results are presented with appropriate caveats, such as the need for further exploration and the specific nature of the dataset.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is moderate. The presentation is based on a published paper (mentioned but not cited in detail), and the dataset is publicly available. However, the methodology is not fully detailed, and the comparison with classical methods may not be exhaustive. The sources are primarily the company’s own materials and the conference presentation. The title accurately reflects the content, and the presentation is well-structured. The speakers are credible, with affiliations to a university and a quantum computing company. The lack of detailed citations and the promotional nature of the talk slightly reduce the overall rigor.

The title accurately reflects the content: a presentation by two speakers from QCI and SMU on quantum-boosted fraud detection.

7/10

Presentation of a specific application of quantum optimization to fraud detection, with concrete results and a published paper, but limited methodological detail and potential conflicts of interest (company presentation).

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction and forward-looking statements
• 00:30Overview of non-convex optimization and EQC technology
• 02:00Explanation of Dirac 3 hybrid implementation and capabilities
• 04:00Comparison of Dirac 3 vs gradient descent in escaping local minima
• 05:00Specifications of Dirac 3: variables, connectivity, power, packaging
• 06:00Path to all-optical EQC and micro ring resonator component
• 07:00Paul Griffin introduces fraud detection problem and dataset
• 09:00Hybrid classical-quantum boosting approach explained
• 11:00Results: AUC ROC and AUCPR improvements, false positives reduction
• 13:00Discussion of productionization and future work, Q&A

• Q2B Conference Website — Conference where this presentation was given, providing context and possibly links to related materials.

• Q2B Conference Website — Official conference page, likely containing abstracts and possibly links to the published paper.