def random_select_seg(train_inputs, train_targets, seg_table):
#print("inputs, targets, seg_table")
#print(train_inputs)
#print(train_targets)
#print(seg_table)
length = len(seg_table)
#print("length:")
#print(length)
import random
seg_index_start = random.randint(0, length - 3)
seg_index_end = random.randint(seg_index_start + 1, length - 1)
input_start = int(seg_table[seg_index_start])
input_end = int(seg_table[seg_index_end])
print("seg_start, seg_end, input_start, input_end")
print(seg_index_start, seg_index_end, input_start, input_end)
# cut the roi out of train_inputs and train_targets
roi_inputs = train_inputs[:, input_start:input_end, :]
#print("roi_inputs")
#print(roi_inputs)
roi_targets = train_targets[seg_index_start:seg_index_end]
#print("roi_targets")
#print(roi_targets)
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets
def hotspot_select_seg(train_inputs, train_targets, seg_table):
length = len(seg_table)
hot_pos = -1
hot_pos_add = find_hotspot(train_targets,add_op)
hot_pos_incept = find_hotspot(train_targets, incept_op)
#print(hot_pos_add)
if(hot_pos_add == -1):
if(hot_pos_incept == -1):
return(random_select_seg(train_inputs,train_targets,seg_table))
else:
hot_pos = hot_pos_incept
else:
hot_pos = hot_pos_add
import random
left_max_range = hot_pos
right_max_range = length - hot_pos
r_left = max(0,left_max_range-70)
#r_right = min()
left_range = random.randint(r_left, left_max_range)
right_range = random.randint(2, right_max_range)
if right_range-left_range>120:
right_range = left_range + random.randint(0,120)
seg_index_start = hot_pos - left_range
seg_index_end = hot_pos + right_range-1
input_start = int(seg_table[seg_index_start])
input_end = int(seg_table[seg_index_end])
#print(seg_index_start, seg_index_end, input_start, input_end)
roi_inputs = train_inputs[:,input_start:input_end,:]
roi_targets = train_targets[seg_index_start:seg_index_end]
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets
def deterministic_select_seg(train_inputs, train_targets, seg_table,
start_index, end_index):
#print("inputs, targets, seg_table")
#print(train_inputs)
#print(train_targets)
#print(seg_table)
print("+++++++++++++++++++++++++++++++++++++here++++++++++++++++")
train_inputs = transform_dd(train_inputs, train_targets, seg_table)
length = len(seg_table)
#print("length:")
#print(length)
#import random
#seg_index_start = random.randint(0,length-3)
seg_index_start = start_index
#r_seg_index_end = min(seg_index_start + 120, length-1) #fixed the length as 100?
seg_index_end = end_index
#seg_index_end = random.randint(seg_index_start+1, r_seg_index_end)
input_start = int(seg_table[seg_index_start])
if seg_index_end == length - 1:
input_end = int(seg_table[seg_index_end]) + 1
else:
input_end = int(seg_table[seg_index_end])
print("seg_start, seg_end, input_start, input_end")
print(seg_index_start, seg_index_end, input_start, input_end)
# cut the roi out of train_inputs and train_targets
roi_inputs = train_inputs[:, input_start:input_end, :]
#print("roi_inputs")
#print(roi_inputs)
roi_targets = train_targets[seg_index_start:seg_index_end]
#print("roi_targets")
#print(roi_targets)
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets
def end_select_seg(train_inputs, train_targets, seg_table):
length = len(seg_table)
range_left = max(0, length-120)
import random
seg_index_start = random.randint(range_left,length-3)
#seg_index_end = random.randint(seg_index_start+1, length-1)
seg_index_end = length-1
input_start = int(seg_table[seg_index_start])
input_end = int(seg_table[seg_index_end])+1
print("seg_start, seg_end, input_start, input_end")
print(seg_index_start,seg_index_end,input_start,input_end)
roi_inputs = train_inputs[:,input_start:input_end,:]
roi_targets = train_targets[seg_index_start:seg_index_end]
#print("roi_targets")
#print(roi_targets)
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets
def hotspot_select_seg(train_inputs, train_targets, seg_table):
length = len(seg_table)
hot_pos = -1
hot_pos_add = find_hotspot(train_targets,add_op)
hot_pos_incept = find_hotspot(train_targets, incept_op)
#print(hot_pos_add)
if(hot_pos_add == -1):
if(hot_pos_incept == -1):
return(random_select_seg(train_inputs,train_targets,seg_table))
else:
hot_pos = hot_pos_incept
else:
hot_pos = hot_pos_add
import random
left_max_range = hot_pos
right_max_range = length - hot_pos
left_range = random.randint(0, left_max_range)
right_range = random.randint(2, right_max_range)
seg_index_start = hot_pos - left_range
seg_index_end = hot_pos + right_range-1
#print(seg_index_start, seg_index_end, input_start, input_end)
roi_inputs = get_input_sequence(train_inputs, seg_index_start, set_index_end, seg_table)
roi_targets = train_targets[seg_index_start:seg_index_end+1]
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets
def random_select_seg(train_inputs, train_targets, seg_table):
#print("inputs, targets, seg_table")
#print(train_inputs)
#print(train_targets)
#print(seg_table)
length = len(seg_table)
#print("length:")
#print(length)
import random
seg_index_start = random.randint(0,length-5)
seg_index_end = random.randint(seg_index_start+1, length-1)
if(seg_index_end - seg_index_start>MAX_RANGE):
if seg_index_end%2 ==0:
seg_index_start = seg_index_end - MAX_RANGE
else:
seg_index_end = seg_index_start + MAX_RANGE
roi_inputs = get_input_sequence(train_inputs, seg_start, seg_end, seg_table)
print("seg_start, seg_end, input_start, input_end")
print(seg_index_start,seg_index_end,input_start,input_end)
# cut the roi out of train_inputs and train_targets
roi_inputs = train_inputs[:,input_start:input_end,:]
#print("roi_inputs")
#print(roi_inputs)
roi_targets = train_targets[seg_index_start:seg_index_end+1]
#print("roi_targets")
#print(roi_targets)
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets
def all_select_seg(train_inputs, train_targets, seg_table):
roi_inputs = train_inputs
roi_targets = train_targets
from data_processor_seg import sparse_tuple_from
roi_targets_sparse = sparse_tuple_from([roi_targets])
return roi_inputs, roi_targets_sparse, roi_targets