QTML 2025: Interactive proofs for verifying (quantum) learning and testing

Value of the Information & Strength of the Argument

The talk provides a clear and rigorous presentation of original research. The argumentation is well-structured, starting with motivation, formal framework, negative results, positive results, and open questions. The value lies in addressing a timely problem in quantum computing: how to delegate learning tasks to untrusted parties with more resources. The negative results are significant, showing limitations of classical interaction, while the positive results with quantum communication and VBQC offer practical pathways. The argumentation is solid, based on formal theorems and proofs, and the speaker effectively explains complex concepts.

Scientific Rigor, Source Quality, Title Accuracy

The talk is based on a paper by the speaker and collaborators, which is a credible source. The presentation is rigorous, with formal statements and references to prior work (e.g., by Sitan Chen and others). The title accurately reflects the content. The talk does not include external sources beyond the paper, but the methodology is sound. The adequacy between title and content is high.

The title accurately reflects the content, focusing on interactive proofs for verifying quantum learning and testing.

8/10

Presentation of original research with formal theorems and proofs, based on a paper by recognized researchers in quantum computing. The talk is technical and assumes background in quantum information, but the content is rigorous and well-structured.

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

• 00:00Introduction and motivation for outsourcing quantum data processing
• 02:00Purity testing example and resource constraints
• 05:00Formal framework: many-versus-one distinguishing tasks and interactive proofs
• 08:00Negative results: classical communication does not help
• 10:00Positive results with quantum communication and VBQC
• 12:00Blackbox interactive proofs and leveraging prover's data access
• 14:00Applications: state tomography and stabilizer state learning
• 16:00Conclusion and open questions

• Paper: Interactive proofs for verifying (quantum) learning and testing — The talk is based on this paper by the speaker and collaborators.

• Sitan Chen et al. - Exponential separations between learning with and without quantum memory — Prior work showing hardness of purity testing with limited quantum memory, which the talk builds upon.