[ИАД, весна 2026] Введение в машинное обучение. Лекция 1: Научный метод и основы машинного обучения

[ИАД, весна 2026] Введение в машинное обучение. Лекция 1: Научный метод и основы машинного обучения

by Machine Learning – Intelligent Systems

Formal & Physical Sciences AI — Models & Research UYQArtificial intelligence (AI)UYQMMachine learning
🎙 Machine Learning – Intelligent Systems 👥 8K 📅 February 12, 2026 ⏱ 109 min 👁 452 📄 lecture 🧭 2026-08-16
Available in: English (current) Français

Keywords

machine learningscientific methodregressionclassificationleast squares

Summary

This introductory lecture on machine learning, part of a spring 2026 course, begins with the instructor emphasizing the importance of face-to-face interaction despite the remote format. The course aims to provide a helicopter view of machine learning, structured differently from the full-year course. The lecture opens with two epigraphs: Klaus Schwab’s 2016 statement about the Fourth Industrial Revolution and a White House report on AI’s impact on automation, highlighting the growing importance of AI and machine learning. The instructor stresses that machine learning is now a mainstream technology, and the course will focus on the mathematical foundations, aiming to train the intellectual elite who build new technologies. The core of the lecture introduces the basic notation: objects (x_i), features (f_j), and target variables (y_i), with two main problem types: regression (numeric target) and classification (discrete target). The instructor emphasizes the importance of defining the problem’s DNA (Given, Find, Criterion) before solving it. He introduces the concept of empirical risk minimization and loss functions. Historical examples illustrate these concepts: Francis Bacon’s ’tables of discovery’ as a precursor to data matrices, Gauss’s use of least squares to determine asteroid orbits (with a discussion of different problem formulations), and Galton’s work on regression towards mediocrity, which introduced the term ‘regression’ and demonstrated a data-driven approach without a physical model. The lecture concludes by connecting these ideas to the scientific method, suggesting that machine learning is a modern embodiment of empirical inquiry.

238 words

Critical Evaluation

Value of the Information & Strength of the Argument

The lecture provides a solid conceptual foundation for machine learning, effectively linking historical developments to modern practice. The instructor’s argumentation is clear and logical, using concrete examples (Gauss’s orbit determination, Galton’s heredity study) to illustrate abstract concepts. The emphasis on problem formulation (DNA) and the distinction between supervised and unsupervised learning are valuable. The discussion of different ways to formalize the same problem (ellipse fitting) highlights the importance of modeling choices. The historical narrative is engaging and helps contextualize the field’s evolution. However, the lecture is introductory and does not delve into technical details or advanced methods, which is appropriate for the target audience.

Scientific Rigor, Source Quality, Title Accuracy

The lecture demonstrates scientific rigor in its presentation of historical facts and mathematical concepts. The instructor accurately describes the contributions of Bacon, Gauss, and Galton, and correctly explains the method of least squares and regression. However, no formal citations are provided within the lecture, and the description contains no links to sources. The title accurately reflects the content, as the lecture indeed covers the scientific method and foundations of machine learning. The instructor’s expertise is evident, but the lack of explicit references to literature or external resources limits the verifiability of the claims.

212 words

Title / Content Match

The title accurately reflects the content: an introductory lecture on machine learning covering the scientific method and foundational concepts.

Quality & Reliability

8/10

The lecture is delivered by an experienced academic (head of department) with over 30 years in the field. It provides a structured introduction to machine learning, referencing historical figures (Bacon, Gauss, Galton) and core concepts. The content is mathematically sound and pedagogically clear, though it lacks formal citations to specific sources.

Key Moments

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

Contribution & Novelties

The lecture provides a unique historical perspective on machine learning, tracing its roots to the scientific method and early statistical methods. It emphasizes the importance of problem formulation and the distinction between supervised and unsupervised learning. The instructor’s pedagogical approach, using interactive questions and multiple formulations of the same problem, is effective for building intuition.

Pour aller plus loin :

  • Francis Bacon — The philosopher who proposed the empirical method, foundational to modern science.
  • Least squares — The method introduced by Gauss and Legendre, central to regression.
  • Regression toward the mean — Galton’s concept, explaining the phenomenon observed in his heredity studies.

102 words

Radar Profile

The radar profile shows high scores in information quantity, quality, and reliability, with a slightly lower technical level, reflecting the introductory nature of the lecture. The balance between historical context and technical foundations is well maintained.

Reliability 8/10