I don't really see how it is so obvious that each step decreases the objective function and, therefore, why it is so clear that based on the results from Q33 and Q34 it must converge. Could you explain this in a bit more detail, please?
Hi,
At each step, both the assignment and calculating the new means try to optimize on the objective function, then after each step, the evaluated objective function will be less or equal than the step before.
Therefore, convergence is guaranteed because we keep decreasing the objective. However, it may converge to a local rather than global minimum.
Exam 2020 Q35
I don't really see how it is so obvious that each step decreases the objective function and, therefore, why it is so clear that based on the results from Q33 and Q34 it must converge. Could you explain this in a bit more detail, please?
1
Hi,
At each step, both the assignment and calculating the new means try to optimize on the objective function, then after each step, the evaluated objective function will be less or equal than the step before.
Therefore, convergence is guaranteed because we keep decreasing the objective. However, it may converge to a local rather than global minimum.
Karim
Add comment