Data Fundamentals (H) - Week 07 Quiz
1. Simulated annealing uses what metaheuristic to help avoid getting trapped in local minima?
Randomised restart.
A temperature schedule.
A population of solutions.
Hill climbing.
Crossover rules.
2. A
hyperparameter
of an optimisation algorithm is:
A value that is used to impose constraints on the solution.
A direction in hyperspace.
A measure of how good a solution is.
The determinant of the Hessian.
A value that affects how a solution is searched for.
3. First-order optimisation requires that objective functions be:
one-dimensional
invertible
monotonic
\(C^1\) continuous
disconcerting
4. The gradient vector \(\nabla L(\theta)\) is a vector which, at any given point \(\theta\) will:
be equal to \(\theta\)
point in the direction of steepest descent
be zero
point towards the global minimum of \(L(\theta)\)
have \(L_2\) norm 1
5. Finite differences is not an effective approach to apply first-order optimisation because:
none of the above
the effect of measurement noise
the curse of dimensionality
all of the above
of numerical roundoff issues.
6. Ant colony optimisation applies which two metaheuristics to improve random local search?
temperature and memory
thants
random restart and hyperdynamics
memory and population
gradient descent and crossover
Submit Quiz