Concerning the latest release HCRF13d.zip. Compiling on Linux.
1) In the makefile of the root directory in line 35 a backslash is missing.
2) At least on my system I needed to add "#include <limits.h>" to include/inferenceengine.h
3) The include flag in the makefile needs to be extended such that the header asa.h can be found
INCLUDE_FLAGS = -Iinclude/ \
-Ioptimizers/cgDescent/include/ \
-Ioptimizers/uncOptim/ \
-Ioptimizers/asa/
4) I furthermore needed to add "-fPIC". Because I have no idea what this is doing I added it to every line of the makefile containing $(CXX)
I finally ended up by a missing file "lib/asa.a". (I am still on it..)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Almost done :-)
Now add $(INCLUDE_FLAGS) to
lib/hCRF.so: $(HCRF_OBJS)
$(CXX) -shared -Wl,-soname,$@ $(HCRF_OBJS) $(INCLUDE_FLAGS) -o $@
and I got no more problems in executing matHCRF. Unfortunately I am still not able to verify the compiled files because the matlab example is not complete. There are no models in sampleData.mat :-(
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
We are about to release version 2.0 beta which should solve all these issues with compilers. It will also add support for Python, Windows/Linux x64 and include new extensions to LDCRF and HCRF.
Stay tuned :)
Lp
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
hi
i think you want the hidden state probeblities for each LDCRF node, given input sequnce?
it is better to edit "matHCRF.cpp" and then compile it agian.
but there is a tricky way without needs to change cpp file.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
That was like a disaster for me to compile on new visual studio, it lasted a whole week!
You should do these matters in order to solve the problem completely: use hCRF_full_3.0a.zip as base open hCRF_full_3.0a\apps\hcrf_sln (not others .sln files) set matHCRF project as defaults in solution for matHCRF u should set properties->linker->command line -> additional options:
/FORCE:MULTIPLE /ignore:4006
* for hCRF set properties->librarian->command line -> addiational options:
/ignore:4006
If the above matters are correctly done, you will be able to compile your project for regular and openMP on vs 2013 and above,u also able to link visual studio to matlab for debugging.
Welcome to Open Discussion
Concerning the latest release HCRF13d.zip. Compiling on Linux.
1) In the makefile of the root directory in line 35 a backslash is missing.
2) At least on my system I needed to add "#include <limits.h>" to include/inferenceengine.h
3) The include flag in the makefile needs to be extended such that the header asa.h can be found
INCLUDE_FLAGS = -Iinclude/ \
-Ioptimizers/cgDescent/include/ \
-Ioptimizers/uncOptim/ \
-Ioptimizers/asa/
4) I furthermore needed to add "-fPIC". Because I have no idea what this is doing I added it to every line of the makefile containing $(CXX)
I finally ended up by a missing file "lib/asa.a". (I am still on it..)
Download ASA package from https://sourceforge.net/projects/asa-caltech/files/
1) Edit ASA-Makefile by adding "-fPIC" to CFLAGS "CFLAGS = $(CDEBUGFLAGS) $(DEFINE_OPTIONS) -fPIC"
2) create a library "ld -G *.o -o libasa.so"
3) Copy new library into HCRF/lib
3) Edit HCRF Makefile and change "lib/asa.a" to "lib/libasa.so"
4) Edit HCRF Makefile and change "lib/hCRF.a" to "lib/hCRF.so"
Almost done :-)
Now add $(INCLUDE_FLAGS) to
lib/hCRF.so: $(HCRF_OBJS)
$(CXX) -shared -Wl,-soname,$@ $(HCRF_OBJS) $(INCLUDE_FLAGS) -o $@
and I got no more problems in executing matHCRF. Unfortunately I am still not able to verify the compiled files because the matlab example is not complete. There are no models in sampleData.mat :-(
Hello! Bonjour!
We are about to release version 2.0 beta which should solve all these issues with compilers. It will also add support for Python, Windows/Linux x64 and include new extensions to LDCRF and HCRF.
Stay tuned :)
Lp
Hi
Does anybody know if is there a way for accessing the hidden states visited by a sequence in an LDCRF model?
Results seem to be limited to label states but hidden states are missing.
Thanks
Hi, did you find an answer to this question? I am also looking into this, i would like to see the hidden states for the hCRF.
Thanks
hi
i think you want the hidden state probeblities for each LDCRF node, given input sequnce?
it is better to edit "matHCRF.cpp" and then compile it agian.
but there is a tricky way without needs to change cpp file.
Ok. So far i am having problems importing and compiling in visual studio 2013. Migration report attached.
That was like a disaster for me to compile on new visual studio, it lasted a whole week!
You should do these matters in order to solve the problem completely:
use hCRF_full_3.0a.zip as base
open hCRF_full_3.0a\apps\hcrf_sln (not others .sln files)
set matHCRF project as defaults in solution
for matHCRF u should set properties->linker->command line -> additional options:
/FORCE:MULTIPLE /ignore:4006
* for hCRF set properties->librarian->command line -> addiational options:
/ignore:4006
If the above matters are correctly done, you will be able to compile your project for regular and openMP on vs 2013 and above,u also able to link visual studio to matlab for debugging.