Data Fundamentals (H) - Week 10 Quiz
1. The Nyquist limit \(f_n\) is equal to:
half the sampling rate \(f_s\)
twice the amplitude quantization levels
twice the sampling rate \(f_s\)
1.0Hz
half the amplitude quantization levels
2. Decreasing the number of levels of amplitude quantization will have what affect on the sampled representation of a signal?
Decreased SNR
No effect
Decreased Nyquist rate
Increased SNR
Frequency shift
3. The exponential smooth is often used instead of a moving average because:
it is super-quadratic
it is nonlinear
it requires storing/computing less data
it is probabilistic
it is more numerically stable
4. Aliasing is caused by sampling signals with:
noise levels greater than the maximum SNR
noise levels less than the maximum SNR
frequencies less than the Nyquist limit
frequencies greater than the Nyquist limit
undefined values present
5. Along with a way to evaluate the likelihood and prior at any parameter setting \(\theta\), what else does Metropolis-Hastings need to sample from the posterior distribution?
The square root of 2.
A maximum likelihood estimation procedure.
A proposal distribution \(q(\theta^\prime|\theta)\)
A way to evaluate the evidence \(P(\theta)[/theta]
An integration function \(V(\theta|D)\)
6. In medical device, if you had an initial heart/pulse rate
p0
and a 1D vector of changes in pulse rates captured at evenly spaced intervals,
delta_p
, how would you compute
p
, the pulse rate at each of these times?
p = p0 + np.cumsum(delta_p)
p = p0 * delta_p
p = p0 + delta_p[:]
p = np.sum(delta_p) + p0
p = np.prod(delta_p) * p0
Submit Quiz