#Sales Offer!| Get upto 25% Off:

1. Consider the mapping φ : R→Rd+1 defined by φ(x) = (x,       x              2). Show that if x1, . . .,xare shattered by Bd then φ(x1), . . .,φ(xm) are shattered by the class of halfspaces in Rd+1 (in this question we assume that sign(0) = 1). What does this tell us about VCdim(Bd)?

2. (*) Find a set of +1 points in Rthat is shattered by Bd . Conclude that +1 ≤ VCdim(Bd ) ≤ +2.

3. Boosting the Confidence: Let be an algorithm that guarantees the following: There exist some constant δ0 ∈ (0,1) and a function mH : (01) → N such that for every ∈ (01), if ≥ mH(_) then for every distribution it holds that with probability of at least 1−δ0, LD(A(S))≤ minhH LD(h)+_. Suggest a procedure that relies on and learns in the usual agnostic PAC learning model and has a sample complexity of mH(_, δ) ≤ kmH(_)+

2log(4k_, where =_log(δ)/log(δ0)_.

Hint: Divide the data into +1 chunks, where each of the first chunks is of size mH(_) examples. Train the first chunks using A. Argue that the probability that for all of these chunks we have LD(A(S))minhH LD(h)+is at most δk0 ≤ δ/2.

Finally, use the last chunk to choose from the hypotheses that generated from the chunks (by relying on Corollary 4.6).

Found something interesting ?

• On-time delivery guarantee
• PhD-level professional writers
• Free Plagiarism Report

• 100% money-back guarantee
• Absolute Privacy & Confidentiality
• High Quality custom-written papers

Related Model Questions

Feel free to peruse our college and university model questions. If any our our assignment tasks interests you, click to place your order. Every paper is written by our professional essay writers from scratch to avoid plagiarism. We guarantee highest quality of work besides delivering your paper on time.

Grab your Discount!

25% Coupon Code: SAVE25
get 25% !!