issues
search
shachen
/
PLDS
The Penalized Linear Dynamical System Project
18
stars
12
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
"In fλ1 (C; {x}, {y}), C is a"
#26
jovo
closed
9 years ago
0
eq 7 swap left and right sides
#25
jovo
closed
9 years ago
0
Estimation for transition matrix C also has closed form solution, but the solution can only be derived by rearranging the terms properly.
#24
jovo
closed
9 years ago
0
align partial derivative equations in M estimation section
#23
jovo
closed
9 years ago
0
The parameters are Θ = {A, C, R, π0}. Each of them is re-estimated
#22
jovo
closed
9 years ago
0
i'd but eq 6 all on 1 line (not a big deal obviously)
#21
jovo
closed
9 years ago
0
when going to eq 4 you dropped terms,
#20
jovo
closed
9 years ago
0
' = transpose?
#19
jovo
closed
9 years ago
0
the definitions of p(y_t | x_t) and p(x_t | x_t-1) are strange
#18
jovo
closed
9 years ago
0
"A new algorithm is then developed and detailed as follows."
#17
jovo
closed
9 years ago
2
"directly applied to model (3) "
#16
jovo
closed
9 years ago
0
"and then refining the solution with EM is a worthwhile approach "
#15
jovo
closed
9 years ago
0
"Specifically, one way to find the maximum likelihood solution is through iterative techniques such as expec- tation maximization (EM) [23]. "
#14
jovo
closed
9 years ago
0
does the MLE even exist for our problem? i think so, but are we sure?
#13
jovo
opened
9 years ago
6
"the exact and variational learning"
#12
jovo
closed
9 years ago
0
"The motivating application requires parameter estimation in model (3)"
#11
jovo
closed
9 years ago
0
(3) vs. (2)
#10
jovo
closed
9 years ago
1
eq. 2 is too informal
#9
jovo
closed
9 years ago
1
"Having those signals spatially smooth allows capturing the active regions within the motor cortex."
#8
jovo
closed
9 years ago
0
when introducing y,x, A,C, Q, and R
#7
jovo
closed
9 years ago
0
abstract goes in 'abstract' environment, it is not a numbered section
#6
jovo
closed
9 years ago
0
the page number overlaps with text, change that.
#5
jovo
closed
9 years ago
0
"Therefore some constraints on R are necessary"
#4
jovo
closed
9 years ago
1
"and too general to be useful."
#3
jovo
closed
9 years ago
0
start general, and then state assumptions
#2
jovo
closed
9 years ago
2
make sure the real experiments get the data properly
#1
jovo
opened
9 years ago
1
Previous