Unofficial Python3 version of "A Laplacian Framework for Option Discovery in Reinforcement Learning" paper.
Second PVF (left) and its corresponding option (right) in the 4-room domain. Action terminate is depicted in red (top right corner), other actions are depicted as arrows.
Options obtained from the four smallest eigenvectors in the 4-room domain. Action terminate is depicted in red.
![]() |
![]() |
![]() |
![]() |
---|
Options obtained from the four smallest eigenvectors in the 4-room domain. Action terminate is depicted using dots.
![]() |
![]() |
![]() |
![]() |
---|
Value functions in 4-room domain.
![]() |
![]() |
![]() |
![]() |
---|
PVFs in 4-roon domain.
![]() |
![]() |
![]() |
![]() |
---|
Options obtained from the four smallest eigenvectors in the 4-room domain. Action terminate is depicted using dots.
![]() |
![]() |
![]() |
![]() |
---|
Value functions in 4-room domain.
![]() |
![]() |
![]() |
![]() |
---|
PVFs in 4-roon domain.