Signal Processing - July 2016 - 66

Signal of the Day Series (EPFL)

(°C)

As we were preparing the third
edition of the École Polytechnique
Fédérale de Lausanne Digital
Signal Processing (DSP) massive
open online course, we thought it
would be cute to start each lecMean Annual Temperature
12
ture with a signal of general inter11
est, tell its history, and show some
10
notable examples of relevant pro9
cessing. We went the extra mile
to try to find "original" signals,
8
either for their historical impor7
tance or with respect to their rele6
1820
1852
1884
1916
1948
1980
2012
vance to concepts taught in the
Year
class. The challenge was then to
transform a cute idea into an
attractive and engaging three- to
FIGURE S1. The Goethe time series.
five-minute video (or an IPython
Notebook) that we call the Signal
of the Day series.
There was no shortage of ideas, and colleagues from the
Table S1. Signals of the day.
lab suggested examples ranging from the mundane to the
downright bizarre; after all, signals are everywhere. Our
von Goethe's Temperature Measurements
The Tristan Chord
goal, in all of this, was to reach a broader community than
Lehman Brothers' stock value ca. 2008
"just" those involved in signal processing; we wanted a
Sputnik: The first man-made signal from outer space
collection of signals from different scientific communities
I Feel Fine: The Beatles and guitar distortion
and those that involved diverse processing challenges. We
Can one hear the shape of a room? [10]
are convinced, after all, that a lot of people in science do
Moiré patterns
Camera resolution and space exploration
signal processing without realizing it; reaching out to these
Exoplanet hunting
communities is both interesting and fun.
Safecast: Crowdsourced radioactivity measurements after Fukushima [9]
As an example, while talking to a German environmental
scientist, we discovered that in 1821, Johann Wolfgang
von Goethe, the famous German writer, had started taking
Currently, we have ten signals of the day, with a few more
daily temperature measurements in his hometown of Jena.
under construction (see Table S1). A fun one is certainly the
This practice has been carried on to the present day by the
one explaining why one should never wear a striped shirt
Jena weather station (with only a few exceptions during
when on television, a playful way to introduce aliasing and
World War II), and so Goethe's time series is probably one
its effects. It was also interesting to discover that other comof the oldest "live" discrete-time records in existence [22].
munities have developed similar ideas (see, for instance, the
We turned this into our first "signal of the day," applying a
"astronomy picture of the day" [23]); as signal processing
simple moving average filter to demonstrate that, despite
moves more and more to the online medium, an ever-growclaims by climate change skeptics, temperature is indeed
ing library of notable DSP examples, with contributions from
rising (Figure S1).
the entire DSP community, would be a fantastic project.

Engagement
For a more dynamic perspective, the engagement data over
time for the fourth edition of the class are shown in Figure 2.
The curves are very similar in shape for all editions and allow
us to draw the following quantitative analysis:
■■ Attendance measured as passive visits to the website displays
a steep decay in the first weeks of class and levels off after
66

■■

week four. This roughly corresponds to the end of the Fourier
analysis module, i.e., those who survive Fourier seem to keep
their interest alive throughout the rest of the course.
The percentage of students who submit homework seems
to remain constant, i.e., from the start the number of students who attempt to obtain a certificate represents a
small percentage of the total number of participants.

IEEE Signal Processing Magazine

July 2016

Table of Contents for the Digital Edition of Signal Processing - July 2016