Noise Filtering

Question 1

______ _______ is removing noise from a signal. There are several ________ algorithms that achieve this

Answer 1

Signal processing, filtering

Question 2

What is LOESS smoothing?

Answer 2

A technique to smooth a curve and to remove the effects of noise

Question 3

What are the steps in LOESS?

Answer 3

Take local section of data
Fit a line to it
Declare that line to be the part of the curve for the middle of the section
Slide the window along, generating a curve as you go

Question 4

In LOESS smoothing, what do small and large fractions imply?

Answer 4

Small fractions are sensitive to noise and a small region

Large fractions wont respond as the signal changes

Question 5

LOESS is best used when?

Answer 5

When there are lots of samples

Question 6

______ filtering is designed to take into account domain knowledge of the signal

Answer 6

Kalman

Question 7

Kalman filtering works best if you know the following 4 things:
How much \_\_\_\_\_\_ in measurements
How \_\_\_\_\_ it changes
A \_\_\_\_\_\_ for the next value
How much \_\_\_\_\_ you expect on the above

Answer 7

error, fast, prediction, error

Question 8

In Kalman filtering, we assume out noise follows a _______ distribution

Answer 8

normal

Question 9

The range from avg-std dev. to avg+std dev. contains __% of the observations

Answer 9

68

Question 10

______ says how 2 variables relate

Answer 10

covariance

Question 11

What are transition_matrix, observation_covariance, and transition_covariance?

Answer 11

transition_matrix: Prediction of next value based on current value
observation_covariance: Represents error in the measurement
transition_covariance: Represents error in your prediction

Question 12

Filtering frequencies is good for ______ data

Answer 12

periodic

Question 13

What is lowpass and highpass?

Answer 13

Lowpass: Keep low and discard high frequencies
Highpass: Opposite of lowpass

Question 14

The _________ filter is a common choice to filter signals

Answer 14

Butterworth