shachen PLDS issues - Githubissues

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

Modify the kalman filtering part to reduce duplicate computation

#76 yangyutu closed 9 years ago
0
update text to reflect new times

#75 jovo closed 9 years ago
3
plot accuracy & time vs. iteration #

#74 jovo opened 9 years ago
5
C'*inv(R)

#73 jovo closed 9 years ago
0
read jovo's fast non-negative paper

#72 jovo closed 9 years ago
0
profile KFS code

#71 jovo closed 9 years ago
0
add everyone's affiliations

#70 jovo closed 9 years ago
3
fig 9

#69 jovo closed 9 years ago
1
i don't quite understand the SVD method

#68 jovo closed 9 years ago
0
"as there is no way that we can predict the two noise terms in the model, "

#67 jovo closed 9 years ago
1
" The prediction accuracies are shown in Figure 8. "

#66 jovo closed 9 years ago
0
The first N = 1000 are picked as train data,

#65 jovo closed 9 years ago
0
(Descriptions of data and citations required here)

#64 jovo closed 9 years ago
0
"As an example, the blue region above accurately matches the DM (dorselmedical) parcel of the five-region parcellation proposed by Nebel MB et al. [18]."

#63 jovo closed 9 years ago
0
Table 2: Similarities Among Estimated A Matrices

#62 jovo closed 9 years ago
1
define Amari distance

#61 jovo closed 9 years ago
0
fig 6

#60 jovo closed 9 years ago
0
denote the A matrix estimation for the first scan of subject one as A11

#59 jovo closed 9 years ago
1
et. al should be et al.

#58 jovo closed 9 years ago
0
"Tuning parameters: λA = 0.00001, λC = 0.00001

#57 jovo closed 9 years ago
0
"The PLDS "

#56 jovo closed 9 years ago
1
"and xt contains the mixing coefficients"

#55 jovo closed 9 years ago
1
"Each column of the C matrix is interpreted as a time-invariant brain network."

#54 jovo closed 9 years ago
2
figure captions are not sufficiently detailed or explanatory

#53 jovo closed 9 years ago
1
in 5.2 "PLDS model"

#52 jovo closed 9 years ago
0
fig 1b

#51 jovo closed 9 years ago
1
As a demonstrate,

#50 jovo closed 9 years ago
0
title of fig 1b has typo

#49 jovo closed 9 years ago
0
Fig 2

#48 jovo closed 9 years ago
0
"In addition, when the penalties are getting larger, the penalized algorithm’s estimations should become worse."

#47 jovo closed 9 years ago
0
"with some proper penalty parameters"

#46 jovo closed 9 years ago
0
The calculation of d(A, B) is essentially a linear assignment

#45 jovo closed 9 years ago
1
d(A,B)

#44 jovo closed 9 years ago
4
some distance measure should be first defined.

#43 jovo closed 9 years ago
1
"The parameter are generated in a similar manner."

#42 jovo closed 9 years ago
1
Then the sample is sorted in ascending order

#41 jovo closed 9 years ago
0
Eigen values is 1 word

#40 jovo closed 9 years ago
0
The A matrix is generated such that the conditional number is no less than some threshold, 50 being used.

#39 jovo closed 9 years ago
0
5.1 Parameter Estimations

#38 jovo closed 9 years ago
0
First, a sparse matrix is used to represent R.

#37 jovo closed 9 years ago
1
The complete EM algorithm for PLDS is addressed as follows.

#36 jovo closed 9 years ago
0
step 0 in Appendix is wrong?

#35 jovo closed 9 years ago
1
Appendix 1: Algorithm Kalman Filter Smoother

#34 jovo closed 9 years ago
0
Expectations (6) are calculated with a Kalman filter/smoother, which is detailed in Appendix 1.

#33 jovo closed 9 years ago
0
EM pseudocode

#32 jovo closed 9 years ago
0
With FISTA and L, matrix A can be update:

#31 jovo closed 9 years ago
0
FISTA is an accelerated version of the Iterative Shrinkage- Threshholding Algorithm (ISTA

#30 jovo closed 9 years ago
0
swap left and right sides of eq 11

#29 jovo closed 9 years ago
0
Last but not least, let’s look at parameter A.

#28 jovo closed 9 years ago
0
consider defining the 'vec' operator

#27 jovo closed 9 years ago
0