Is there then a logic to how many layers you need to add in a CNN?
So a filter in the image case (which has three channels) could be a tensor? E.g. a 3 x 3 x 3 tensor?
Sorry, we were expecting the forums to be working.
yes 
1 Like
Since like a minute ago things seem to work in forum
How do you visualise the convolutions (like in the Zeiler paper)?
1 Like
It is a research question. In general which architecture to use it is a big research question.
1 Like
Yes, it is
Yeah, not sure what changed, but things seem to be running smooth on the forum now.
With the 3 most important tools and jupityr being 3rd. What are the first two?
It is not straight forward to explain.
Seems like they used CNN to assess board positions.
See a paper in Nature.
"We pass in the board position as a 19×19 image and use convolutional layers to construct a representation of the position. "
2 Likes
@zaoyang If I’m not wrong, AlphaGo uses CNN in its primary stage to get raw images of board positions which then in is fed into its deep RL and MCTS to predict best move.
1 Like
@yinterian I am not able to find my name in the spreadsheet which Jeremy shared, can you please help.
Thanks
What do you actually need? Since you have access to forums.
I mean in that team list, my name is missing. https://docs.google.com/spreadsheets/d/1MjY5tVIZy5VtBaDSqMI3wURKEWJQQEqYmnvIiIwcCdY/edit#gid=0
1 Like
Does this work on crestle ? I get below error:
In file included from …/ncurses/curses.priv.h:283:0,
from …/ncurses/lib_gen.c:19:
_24926.c:835:15: error: expected ‘)’ before ‘int’
…/include/curses.h:1594:56: note: in definition of macro ‘mouse_trafo’
#define mouse_trafo(y,x,to_screen) wmouse_trafo(stdscr,y,x,to_screen)
^
Makefile:785: recipe for target ‘…/objects/lib_gen.o’ failed
make[1]: *** […/objects/lib_gen.o] Error 1
make[1]: Leaving directory '/home/nbuser/local/tmux_tmp/ncurses-5.9/ncurses’
Makefile:109: recipe for target ‘all’ failed
make: *** [all] Error 2
nbuser@jupyter:~/76a29560d35942558375778abe07fc74$