Quantum Machine Learning Workshop - Day 2 (10.10.2026)

Value of the Information & Strength of the Argument

The video provides a practical, hands-on approach to integrating quantum models into classical deep learning pipelines. It clearly explains key concepts like data encoding, leakage, and class imbalance, using concrete examples and code. The instructor’s argumentation is built on common pitfalls and immediate applicability, which makes the content accessible yet technically thorough. He openly admits to committing bad practice (global preprocessing) to illustrate the risk, which strengthens the pedagogical value. The logic is sound: from defining classification, to preparing data, to implementing and training a hybrid model, each step is justified.

Scientific Rigor, Source Quality, Title Accuracy

The content is rigorous in its technical explanations and reflects established knowledge in quantum machine learning, though it lacks formal citations. The speaker references only his own GitHub resources, providing no external literature or sources. The title corresponds exactly to the workshop’s scope. The discussion on data leakage and class imbalance is well-founded and aligns with standard machine learning guidelines. However, the absence of explicit references to peer-reviewed work limits the scientific robustness. Overall, the technical accuracy is high, but the sourcing is weak, and the audience would benefit from additional reading materials.

Title accurately reflects the content: a second-day workshop on quantum machine learning, covering classification and quantum-PyTorch integration.

7/10

Solid workshop content with clear explanations, but relies on unverified claims and lacks external references. The presented techniques are established in the field, and the practical examples are instructive, yet the absence of citations lowers the reliability.

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

• 00:00Welcome and logistics; mention of final test and feedback form.
• 05:00Recap of last week's problems: sine fitting and encoding conflicts; Swiss roll with re-uploading.
• 15:00Introduction to classification; classical vs quantum classifiers.
• 25:00Data preparation and leakage: global preprocessing pitfalls, feature selection, dimensionality reduction.
• 35:00Class imbalance: accuracy misuse, sensitivity/specificity, Kappa, SMOTE, prior probability correction.
• 45:00Integrating PennyLane with PyTorch: wrapping quantum layer as a PyTorch layer, training loop identical to classical.
• 1:00:00Hands-on code walkthrough: building hybrid models, cost functions, optimizers.
• 1:15:00Hyperparameter tuning with grid search; summary of workshop.