Exam - 2019 Flashcards
What is a LIDAR? What does it measure, how does it work and where is it
used? [5p]
LIDAR is the abbreviation of Light Detection and Ranging. It is a sensor that
sends out a laser beam and it measures the time needed for the laser beam to
come back.
Depending on this time and the wave length of the laser, the sensor
calculates the distance to an obstacle.
Distance = speed x time / 2.
It is used in
self-driving cars to map the surrounding area.
Explain how a mean filter works for 2D-images. Explain how a median
filter works for 2D-images. Give in both cases a small example. [6p]
In median filtering, for EACH pixel in the image, one changes its colour by
calculating the average of its current colour and the colours of all its 8
neighbors. (mean filtering)
For median filtering, we do the same, but instead of calculating the average, we
order all the colours in ascending order, and the new colour of the pixel in the
center will be the median value of this sequence.
Define the term spectrogram and explain how it is constructed.
If a signal is not periodic, then it is useful to use a spectrogram to analyze its
spectrum. The signal has first to be sampled. We divide the signal in time
windows where we can assume that the signal is periodic and then we apply
FFT to the samples from that window. The windows have to overlap, usually
50%. The result of this FFT is called short term Fourier transform (STFT).
Each
STFT is plotted then in time in a 3D graph, called spectrogram. In a
spectrogram one axis is time, another axis is frequency, and the third axis is
the intensity, given by the amplitude of each frequency component.
Intensity is
represented using colour, where a more intense colour means a higher
amplitude.
Imagine we want to control a wheelchair using speech. Can you suggest a
method to solve the recognition problem? Enumerate the steps you need
to take, specify which kind of classifier are you going to use and which
features.
What are the advantages and disadvantages of your approach ?
[6p]
Here many answers are good. For example a rule based classifier using
the FFT spectrum as features, a neural network using MFCC coefficients
as inputs or template matching.
What are Hidden Markov Models (HMM)? Give an example of how to use
HMM in pervasive computing. [8p]
HMM are Hidden Markov models.
These are graphical
probabilistic reasoning algorithms that can be used for example for
classification.
They are dynamic models because the states are evolving
in time.
They contain hidden states and observable states connected with
edges.
Each transition from on estate into another has a certain
probability.
The challenge is to infer the hidden states knowing the visible
observable states.
For example we mount sensors in a house (door, bed,
floor, toilet) in order to monitor its inhabitants daily activity(cooking,
sleeping, washing, toileting, etc).
A HMM will connect the visible states z
given by the sensor readings and the hidden states x (cooking, sleeping,
washing, toileting, etc).
The transition probabilities p(zt|zt-1), p(zt|xt)
are established by long days of training, annotation, log books, etc.
The
hidden states given by the probability that you are in state x knowing the
observable z, p(x|z) are inferred by using Bayes theory.
What is a confusion matrix ? Name and define its components.
Show a
“good” confusion matrix for a (red-yellow-green) traffic light recognition
system used in a smart car, tested on 10 000 traffic lights. Justify your
answer. [6p]
A confusion matrix summarizes the results of a classification test. It gives information on how frequently instances of class A were correctly classified as class A or misclassified as another class.
The entries
in a confusion matrix are counts, i.e. integers, showing the true positives,
the true negatives, the false positive sand the false negatives.
For a
system that recognizes traffic lights, an ideal matrix should have high
values on the diagonal and very low values for the rest.
Imagine you have to build an automatic insulin pump, a wearable device
that keeps the blood glucose concentration BGC of a diabetes patient in a
healthy range of 4-5.5 mmol/L. Insulin is injected any time the BCG is higher
than this range. A BCG lower than this range can be fatal for the patient. The
system stores in a log file all the insulin quantities that have been injected.
Enumerate six stakeholders in this project. [2p]
Diabetes patients, caretakers, families, the manufacturers of the product, the
government, insurance companies, etc
Imagine you have to build an automatic insulin pump, a wearable device
that keeps the blood glucose concentration BGC of a diabetes patient in a
healthy range of 4-5.5 mmol/L. Insulin is injected any time the BCG is higher
than this range. A BCG lower than this range can be fatal for the patient. The
system stores in a log file all the insulin quantities that have been injected.
Formulate one functional requirement and one non-functional
requirement for this system. [4p]
Functional requirement:
REQ1. The system shall maintain the BGC level between 4 and 5.5 mmol/l.
Non functional requirement:
RQ2. The system shall bring the BGC value in the range in maximum 3
seconds.
Or REQ3: The system shall deliver the correct amount of insulin with a 99%
accuracy.
Imagine you have to build an automatic insulin pump, a wearable device
that keeps the blood glucose concentration BGC of a diabetes patient in a
healthy range of 4-5.5 mmol/L. Insulin is injected any time the BCG is higher
than this range. A BCG lower than this range can be fatal for the patient. The
system stores in a log file all the insulin quantities that have been injected.
Identify one accident and for that accident identify one hazard, its
possible causal scenario and specify a way to prevent the hazard from
happening. [6p]
Accident: An insulin pump user dies.
Hazard: Insulin overdose → Unacceptable risk
Causal scenario: the controller has been hacked.
Mitigation measure: Better encryption
Imagine you have to build an automatic insulin pump, a wearable device
that keeps the blood glucose concentration BGC of a diabetes patient in a
healthy range of 4-5.5 mmol/L. Insulin is injected any time the BCG is higher
than this range. A BCG lower than this range can be fatal for the patient. The
system stores in a log file all the insulin quantities that have been injected.
Identify one ethical question relevant to this system. [3p]
Who is responsible for an overdose that can has led to the patient’s death?
Will any patient worldwide have free access to this pump, even if it is
very expensive?
Who will pay the costs? Is this a device only for the rich?
