Work Hours
Everyday: 北京时间8:00 - 23:59
CS 446 / ECE 449 — Homework 1
your NetID here
Version 1.3
Instructions.
- Homework is due Tuesday, February 16, at noon CST; no late homework accepted.
- Everyone must submit individually at gradescope under hw1 and hw1code.
- The “written” submission at hw1 must be typed, and submitted in any format gradescope accepts
(to be safe, submit a PDF). You may use LATEX, markdown, google docs, MS word, whatever you like;
but it must be typed! - When submitting at hw1, gradescope will ask you to mark out boxes around each of your answers;
please do this precisely! - Please make sure your NetID is clear and large on the first page of the homework.
- Your solution must be written in your own words. Please see the course webpage for full academic
integrity information. Briefly, you may have high-level discussions with at most 3 classmates, whose
NetIDs you should place on the first page of your solutions, and you should cite any external reference
you use; despite all this, your solution must be written in your own words. - We reserve the right to reduce the auto-graded score for hw1code if we detect funny business (e.g., your
solution lacks any algorithm and hard-codes answers you obtained from someone else, or simply via
trial-and-error with the autograder). - When submitting to hw1code, only upload hw1.py and hw1 utils.py. Additional files will be ignored.
Version History.
- Initial version.
1.1 Fixed typo in hint for 1c.
1.2 Clarified wording of 1a, 1b.
1.3 Fixed bolding of vectors in 2c.
1 - Linear Regression/SVD.
Throughout this problem let X be the n × d matrix with the feature vectors (xi)
n
i=1 as its rows. Suppose
we have the singular value decomposition X =
Pr
i=1 siuiv
>
i
.
(a) Let the training examples (xi)
n
i=1 be the standard basis vectors ei of R
d with each ei repeated ni > 0
times having labels
