Comment for Final Examination
Regrading Policy for the midterm exam
- Regrading request will be accepted in the
format of resubmission of graded finals until 1pm, Thursday, December 19,
2002. No exception.
- When you resubmit, please accompany extra sheets
of paper where you type out reasons for regrading for specific quesions.
No exception. (hand written request will be ignored.)
- Submit it to the TA mail box or to DEN.
- Please DO NOT contact TA or the professor in
person or by email for regrading; the message
will be ignored.
- Only for obvious mistake of the grader such
as miscounting lost points.
- Only when you have explicitly stated the answers
correctly but still lost point >= 5. (Please DO NOT argue "implied answers,
etc.".)
- Please don't come for point loss less than
5
- Regrading will be done on the entire exam, which
may reduce your grade. No exception.
- No argument against the grading standard given
below.
Grading Standard for the final exam
1
d
genetic algorithm or random starting point (2)
e
depth - disparity; motion - aperture; object - view; visual - inform; scene
- integrative (1)
f
PPC (2)
g
The origin moves with the first saccade (5)
h
look up when needed (3)
reason (1)
2
a
self-organization.(1)
one eye repels its own and attract the others.(2)
resource competition geographically splits neurons with simultaneous firing
apart(3)
b
all columns in a hypercolum correspond to one visual field.(2)
each column processes different cue.(2)
cooperation and competition(2)
easy integration(2)
c
resource/work competition (3)
d
correct explanation of Ferromagnetism (4)
3
a
GEON definition (1)
Recognition by component (1)
Each property (1)
b
edges (3)
c
graphs (2)
missing matrix info (2)
boy&girl (indistinguishable) (2)
d
curvature of each facial feature isolated in a GEON (3)
relative sizes (3)
relative distribution of facial features isolated in a GEON (3)
e
specific (2)
general with learning (2)
general (1)
4
a
first order derivative (2)
second order derivative or inflection point (3)
b
noise reduction (3)
c
noise reduction and edge detection in one step. (3)
Gaussian and Laplacian in one step (restatement of the term itself) (2)
Two step to one step (2)
d
builtin population of filters with various scales (3)
e
a good example (3)
5
a one correct label (1)
b
i
Critic (2)
A or whatever you marked in that box (2)
ii
availability of direction and possibly amount to change (3)
answer reversed (1)
iii
yes (1)
meaning(2)
no(3)
c
i
additional/other feedback (3)
ii
assosiative (3)
iii
dashed input (3)
some good qualitative input (3)