3. Continuing our \"advertisement\" example as discussed in the class. An ad in

Question

3. Continuing our "advertisement" example as discussed in the class. An ad in the "Machine Learning Daily" for a Learning Machine advertises for a machine with the following properties Property 1: Can learn k monotone disjunctions (disjunctions with at most k attributes) from data, where each example is defined using d binary attributes Property 2: Finds a hypothesis which has zero training error. Property 3: 9 times out of 10, learns a hypothesis which will not make more than 5 mistakes in a test data set of 100 randomly picked examples With these properties, the advertisement makes the following claims about the algorithm: . Claim 1: For a 10 dimensional data set, the learner needs less than . Claim 2: For a 10 dimensional data set, the learner needs less than . Claim 3: For a 20 dimensional data set, the learner needs less than . Claim 4: For a 10 dimensional data set, the learner will need less 100 training examples to learn a 5 monotone disjunction. 176 training examples to learn a 10 monotone disjunction. 300 training examples to learm a 20 monotone disjunction. than 180 training examples to learn a 5 monotone disjunction. Which of the following will be true? O a) Both claims 1 and 2 are correct O b) Claim 3 is incorrect O c) Claim 3 is correct O d) Claim 1 is correct and Claim 2 is incorrec

Explanation / Answer

a) Both Claim 1 and 2 are correct.

Because in Claim 1 the learner needs less than 100 training examples to learn 5 monotone diajunction for 10 dimensional data set where as in Claim 2 the matter is same as Claim 1 which includes less than 176 training examples to learn 10 monotone disjunction. So both the Claims are same.

Related Questions

Navigate

• Browse (All)
• Browse 3
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.