Name : Viona Z. A. Kaleb
ID : 001201400064
Class : CIT-2 2014
Machine Learning
Questions
1. Describe in detail, how do we choose weight, using Weight Tuning Rule?
2. Re-draw 2 figures of “The Inductive System” and “The Equivalent Deductive System”
then describe in detail about 2 figures above.
3. Describe in detail how we can avoid over-fitting in Decision Tree?
4. Describe in detail about “The Back-propagation Algorithm” in the Machine Learning.
5. What are they in below: ED(H) ≡ Pr[ f(x) ≠ h(x) ] and Es(h) ≡ ⅟n ∑X∑S ∂( f(x) ≠ h(x) ).
Describe in detail about the definition of them above.
6. Describe in detail about EM Algorithm. What is EM?
7. Describe in detail about Agnostic Learning?
8. Describe in detail about Learning Process to Predict Probabilities.
9. Describe in detail about Radial Base Function Network?
10. Describe in detail about Central Limit Theorem.
Submit to:
Question Answers Which I choose
1. In order to choose and update weight (in V^) using Weight Tuning Rule, we can use one
algorithm which is the Least Mean Squares (LMS) rule.
The LMS Weight update rules are: To select a training example b at random. Then to
compute error (b) : error (b)=V train (b)-V ^(b). After that, for each board feature f i , we
can update weight w i : w i ←w i +c⋅f i ⋅error(b) where, c is some small constant, let’s
say 0.1, in order to moderate the rate of learning. And finally after the algorithm done, we
go back to the first algorithm rule.
ID : 001201400064
Class : CIT-2 2014
Machine Learning
Questions
1. Describe in detail, how do we choose weight, using Weight Tuning Rule?
2. Re-draw 2 figures of “The Inductive System” and “The Equivalent Deductive System”
then describe in detail about 2 figures above.
3. Describe in detail how we can avoid over-fitting in Decision Tree?
4. Describe in detail about “The Back-propagation Algorithm” in the Machine Learning.
5. What are they in below: ED(H) ≡ Pr[ f(x) ≠ h(x) ] and Es(h) ≡ ⅟n ∑X∑S ∂( f(x) ≠ h(x) ).
Describe in detail about the definition of them above.
6. Describe in detail about EM Algorithm. What is EM?
7. Describe in detail about Agnostic Learning?
8. Describe in detail about Learning Process to Predict Probabilities.
9. Describe in detail about Radial Base Function Network?
10. Describe in detail about Central Limit Theorem.
Submit to:
Question Answers Which I choose
1. In order to choose and update weight (in V^) using Weight Tuning Rule, we can use one
algorithm which is the Least Mean Squares (LMS) rule.
The LMS Weight update rules are: To select a training example b at random. Then to
compute error (b) : error (b)=V train (b)-V ^(b). After that, for each board feature f i , we
can update weight w i : w i ←w i +c⋅f i ⋅error(b) where, c is some small constant, let’s
say 0.1, in order to moderate the rate of learning. And finally after the algorithm done, we
go back to the first algorithm rule.
