Week 09 · Lecture 1 · PDF with keyframes

Lecture 1 · Sources and credits

Serega artwork is reused from Albert Nasybullin’s Deep Learning for Search course. Scientific plots, interactive scenes and diagrams were prepared for Intro to ML. Constructed signals and illustrative EEG filters are labeled separately from measured SKAB results.

SKAB: Katser and Kozitsin (2020), dataset and testbed, GPL-3.0 licence. Local fonts are distributed under the SIL Open Font License: IBM Plex Sans, JetBrains Mono, Newsreader, Patrick Hand and Pangolin.

Figure provenance and synthetic-signal conventions.

3. One physical system, eight channels

4. Keep the whole abnormal state visible

6. What does one row mean?

7. One row is not always one second

8. What did RMS erase?

9. Same values. A different story.

10. How far does memory reach?

11. Trend, seasonality, and residuals

13. Stationarity is an assumption to inspect

17. Build a signal from simple waves

18. A repeated shape creates harmonics

19. A Fourier coefficient measures agreement

20. Keep the phase

21. Put units on the frequency axis

25. A wavelet stretches with the question

26. Choose or design the wavelet for the event

27. STFT and CWT ask different local questions

33. Spectral features summarize rhythm

39. Overlapping windows are not independent

40. Roll the forecast origin forward

41. Hold out the identity you expect to meet

42. Guard the information boundary

45. Sources, credits, and reproducible controls