|
| def | __init__ (self, states_df, score_type, max_num_mtries, num_starts=10, ess=1.0, verbose=False, vtx_to_states=None) |
| |
| def | restart (self, mtry_num) |
| |
| def | cache_this (self, move, score_change) |
| |
| def | empty_cache (self) |
| |
| def | __init__ (self, states_df, score_type, max_num_mtries, ess=1.0, verbose=False, vtx_to_states=None) |
| |
| def | climb (self) |
| |
| def | do_move (self, move, score_change, do_finish=True) |
| |
| def | refresh_nx_graph (self) |
| |
| def | would_create_cycle (self, move) |
| |
| def | move_approved (self, move) |
| |
| def | finish_do_move (self, move) |
| |
| def | restart (self, mtry_num) |
| |
| def | cache_this (self, move, score_change) |
| |
| def | empty_cache (self) |
| |
| def | __init__ (self, is_quantum, states_df, vtx_to_states=None) |
| |
| def | fill_bnet_with_parents (self, vtx_to_parents) |
| |
HC_RandRestartLner (Hill Climbing Random Restart Learner) adds to its
parent class HillClimbinmgLner a restart() function for random restarts.
Once a local maximum is reached, a random move ('restart') is made and
then from that new starting point the system is again driven towards a
possibly new local maximum. One can specify the number of random starts
desired. The local maximum with the highest score is selected at the end.
References
----------
1. Nicholas Cullen neuroBN at github
Attributes
----------
is_quantum : bool
True for quantum bnets amd False for classical bnets
bnet : BayesNet
a BayesNet in which we store what is learned
states_df : pandas.DataFrame
a Pandas DataFrame with training data. column = node and row =
sample. Each row/sample gives the state of the col/node.
ord_nodes : list[DirectedNode]
a list of DirectedNode's named and in the same order as the column
labels of self.states_df.
max_num_mtries : int
maximum number of move tries
nx_graph : networkx.DiGraph
a networkx directed graph used to store arrows
score_type : str
score type, either 'LL', 'BIC, 'AIC', 'BDEU' or 'K2'
scorer : NetStrucScorer
object of NetStrucScorer class that keeps a running record of scores
verbose : bool
True for this prints a running commentary to console
vertices : list[str]
list of vertices (node names). Same as states_df.columns
vtx_to_parents : dict[str, list[str]]
dictionary mapping each vertex to a list of its parents's names
mcache : list[tuple[str, str, str]]
a list that stores all moves of a try. This list is cleared at
beginning of each try.
score_ch_cache : list[tuple[float, float, float]]
A list that stores the score changes of each move of a try. Score
changes here are given as a 3-tuple of 3 floats, (beg_score_ch,
end_score_ch, tot_score_ch).
best_start_graph : networkx.DiGraph
best_start_score : float
Each time restart() function is called, the current tot score is
compared with best_start_score, and if it is higher, then both the
current nx_graph and its score are stored in best_start_graph and
best_start_score.
cur_start : int
current start. Increases by one every time restart() is called.
num_starts : int
number of starts. Number of times minus 1 that restart function is
called.
| def learning.HC_RandRestartLner.HC_RandRestartLner.restart |
( |
|
self, |
|
|
|
mtry_num |
|
) |
| |
This function takes in mtry_num and returns (restart_approved,
0). Zero is intended to be the next mtry_num, so mtry_num is set to
zero. Also the cur_start num is increased by one. All restarts are
approved as long as they don't exceed num_starts in number.
The cur total score is compared with best_start_score and it is
higher, then current nx_graph and its score are stored as best so far.
The function selects uniformly at random a move from mcache (mcache
is a list of all non-optimal moves of last try, and score_ch_cache
is a list of the score changes of those non-optimal moves). This
random move is submitted to do_move(), so it is performed regardless
of its non-optimality. The hope is that this random restart will
carry the system from the attractor basin of the current local max,
to the attractor basin of another local max.
Parameters
---------
mtry_num : int
Returns
-------
bool, int
Public Member Functions inherited from learning.HillClimbingLner.HillClimbingLner
1.8.11