Confusing use of gradient in Pattern recognition and machine learning -
i'm reading prml , gradient notation seems confusing. in chapter 2, page 116, a column vector:
and on appendix e page 707, column vector: however, in chapter 3, during derivation of least-square, page 141, row vector:
can clarify these confusing details me? have read posts on web, of them says gradient strictly column vector, says depends on calculation being carried out, says depends on author, , couldn't come conclusive answer
the answer stated in question - gradient vector of partial derivatives, treating column/row vector not matter. people use orientation best particular use/derivation, same applies putting data points in data matrix row/column wise, defining linear projections column/row vectors etc. thus, answer it depends on particular use, have check kind of notation author uses. why that? because not matter, , calculations - row notation reduces amount of transpose operations needed, , - column notation helps. that's all.
Comments
Post a Comment