import Generate
import numpy
import copy
num_feature = 4
hypo_table = Generate.Boundary_Hypo_Table(num_feature)
num_hypo = len(hypo_table)
x = numpy.array(range(num_feature))
for a in range(num_feature):
t = numpy.random.randint(0, num_feature)
x[t], x[a] = x[a], x[t]
print(x)
observation_steps = 2 # How many observations we want
k_matrix = numpy.zeros((num_hypo, num_hypo))
for tr in range(num_hypo):
'''True hypothesis'''
true_hypo = hypo_table[tr]
print(true_hypo)
temp = list(range(num_hypo))
print("temp", temp)
for idx in range(observation_steps):
'''Search for the matching'''
c_idx = x[idx] # The current feature index
c_label = true_hypo[c_idx]
for i in range(num_hypo):
if c_label != hypo_table[i][c_idx]:
# Tree Model
import numpy
import Utility
import copy
import Generate
# Generate the hypothesis matrix
total_features = 4
hypo = Generate.Boundary_Hypo_Table(total_features, True)
total_hypos = len(hypo)
# Create the observation list
arr = ""
for x in range(total_features):
arr += str(x)
obs_list = numpy.array(list(Utility.Permutation(arr)), dtype=int)
lst_size = len(obs_list)
print(obs_list)
best_route = {}
# Count how many observations
counting = 0
for true_idx in range(total_hypos):
# Create the true hypo
true_hypo = hypo[true_idx]
print("True hypothesis = ", true_hypo)
maximum = 0
import AL import Generate import Const #hypo = Generate.Uniform_Hypo_Table(1, False) ''' task = AL.ActiveLearning(knowledgeability=1) Generate.Transfer_User_Table(Const.user_hypo_table, Const.label_map) print(Const.user_hypo_table)P task.Set(user_hypo=Const.user_hypo_table) task.O_Task() ''' task = AL.ActiveLearning(knowledgeability=1) task.Set(user_hypo=Generate.Boundary_Hypo_Table(4, True)) task.DS_Task()
import Generate import numpy import tensorflow import time import copy # =============================== # ====== [Hyperparameters] ====== # =============================== num_feature = 20 num_label = 2 knowledgeability = 1 iteration = 100 hypo_matrix = Generate.Boundary_Hypo_Table(num_feature, True) num_hypo = len(hypo_matrix) ptxy = 1 / num_feature / num_label # =============================== # ====== [Numpy Matrix] ========= # ====== [Memory Usage Note] ==== # ====== [1000 Features] ======== # ====== [About 300 MB RAM] ===== # =============================== # PYXH Matrix P_y_xh = numpy.empty((num_label, num_feature, num_hypo), dtype="float32") # Knowledgeability Matrix Delta_g_h = numpy.zeros((num_hypo, num_hypo), dtype="float32")
