Q1e
http://adv-ml-2017.wikidot.com/forum/t-2279742/q1e
Posts in the discussion thread "Q1e"Tue, 25 Jan 2022 16:39:32 +0000http://adv-ml-2017.wikidot.com/forum/t-2279742#post-2827341Re: Q1e
http://adv-ml-2017.wikidot.com/forum/t-2279742/q1e#post-2827341
Sat, 20 May 2017 14:34:47 +0000Ido Hadanny3026643
Can we assume that B == C? what does it even mean for theta(1,0) != theta(0,1)?
]]>
http://adv-ml-2017.wikidot.com/forum/t-2279742#post-2824254(no title)
http://adv-ml-2017.wikidot.com/forum/t-2279742/q1e#post-2824254
Wed, 17 May 2017 15:19:31 +0000Tom
Are the singletons also the same for every i?
]]>
http://adv-ml-2017.wikidot.com/forum/t-2279742#post-2824091Re: Q1e
http://adv-ml-2017.wikidot.com/forum/t-2279742/q1e#post-2824091
Wed, 17 May 2017 12:37:46 +0000Amir Globerson2872628
I don't mean arbitrary linear transformations. For example consider the transformation of subtracting a constant from A,B,C,D. This doesn't change the maximizer. Similarly if you reduce a rank-one matrix and add it to the singletons it will not change anything (that's a hint).
]]>
http://adv-ml-2017.wikidot.com/forum/t-2279742#post-2821436Q1e
http://adv-ml-2017.wikidot.com/forum/t-2279742/q1e#post-2821436
Mon, 15 May 2017 11:46:08 +0000Tom
Hi,

How is transforming the teta matrix into a matrix that has zeros in every cell but the bottom right doesn't cause data loss on the distribution? Since we consider only p(1,1) for every ij in E when maximizing.