'Skip if already calculated and saved before', action="store_true")
parser.add_argument(
'--plot_prob',
help=
'Plot the graphs of probability of choosing each image vs. fixations',
action="store_true")
parser.add_argument(
'--plot_acc',
help=
'Plot the graphs of prediction accuracy of the network vs. fixations',
action="store_true")
args = parser.parse_args()
# Load data
load_data = LoadData()
slide_sequences, slide_selections = load_data.get_RNN_data_selection()
max_sequence_lengths = load_data.get_max_sequence_lengths()
NUM_SUBJECTS = slide_selections.shape[1]
_, _, _, corr_imgs = load_data.get_other_var()
if args.from_scratch:
#Load models
network = Network()
network.load_models("SELECTION")
models = network.get_models(
) #NUM_SLIDES x k model (each slide has k folds)
stratified_kfold = StratifiedKFold(n_splits=k,
shuffle=True,
random_state=seed)
from utils.ablation_utils import modify_all_sequences
from baseline_model.baseline_model import BaselineModel
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
plt.ion()
from sklearn.model_selection import StratifiedKFold
from keras.preprocessing.sequence import pad_sequences
seed = 7
model_type_name = 'SELECTION'
load_data = LoadData()
slide_sequences, slide_selections = load_data.get_RNN_data() if (
model_type_name == 'CORRECT') else load_data.get_RNN_data_selection()
##### #####
discard_last_dup_sequences = modify_all_sequences(slide_sequences,
num_last_arg=None,
mode='discard last dup',
section=None)
print('slide_sequences[0][0]: ', slide_sequences[0][0])
print('discard_last_dup_sequences[0][0]: ', discard_last_dup_sequences[0][0])
print('slide_sequences[9][9]: ', slide_sequences[9][9])
print('discard_last_dup_sequences[9][9]: ', discard_last_dup_sequences[9][9])
slide_sequences = discard_last_dup_sequences
##### #####
'Skip if already have network trained and saved prior', action="store_true")
parser.add_argument('--conf',
help='Show confusion matrices',
action="store_true")
parser.add_argument('--acc', help='Show accuracies', action="store_true")
parser.add_argument('--auc', help='Show auc_scores', action="store_true")
parser.add_argument('--plot_val_loss',
help='Plot validation losses',
action="store_true")
parser.add_argument('--output',
help='Output dataframes to csv files',
action="store_true")
args = parser.parse_args()
load_data = LoadData()
slide_sequences, slide_labels = load_data.get_RNN_data()
network = Network()
if args.train:
network.train_and_save(slide_sequences,
slide_labels,
"CORRECT",
should_save_auc=True)
else:
network.load_results("CORRECT")
if args.conf:
print('DISPLAYING CONFUSION MATRICES:\n')
conf = network.get_confusion_matrices()
help_method = "1. Shuffle whole sequence\n"+\
"2. Keep only 1st half\n"+\
"3. Keep only 2nd half\n"+\
"4. Shuffle 1st half and keep 2nd half\n"+\
"5. Shuffle 2nd half and keep 1st half\n"+\
"6. Discard 1st half and shuffle 2nd half\n"+\
"7. Discard last 5 fixations"
parser.add_argument('method', choices=range(1,8), help=help_method, type=int)
parser.add_argument('--conf', help='Show confusion matrices', action="store_true")
parser.add_argument('--acc', help='Show accuracies', action="store_true")
parser.add_argument('--auc', help='Show auc_scores', action="store_true")
model_type_name = "CORRECT"
print("model_type_name: ", model_type_name)
load_data = LoadData()
if model_type_name == "SELECTION":
slide_sequences, slide_labels = load_data.get_RNN_data_selection()
else:
slide_sequences, slide_labels = load_data.get_RNN_data()
max_sequence_lengths = load_data.get_max_sequence_lengths()
args = parser.parse_args()
network = Network()
network.load_models(model_type_name)
model = network.get_models()
method = int(args.method)
args = parser.parse_args()
num_features = args.num_features
feature = None
if num_features == '1c':
num_features = 1
feature = 'correct'
elif num_features == '1m':
num_features = 1
feature = 'master'
elif num_features == '1l':
num_features = 1
feature = 'length'
else:
num_features = int(num_features)
load_data = LoadData()
slide_sequences, slide_selections = load_data.get_RNN_data_selection()
num_incorrect, easy_slides, hard_slides, correct_images = load_data.get_other_var(
)
baseline_model = BaselineModel()
baseline_model.train(slide_sequences, slide_selections, correct_images,
num_features, feature)
if args.conf:
print('DISPLAYING CONFUSION MATRICES:\n')
conf = baseline_model.get_confusion_matrices()
for slide_id in range(NUM_SLIDES):
print('Slide ' + str(slide_id + 1) + ':\n', conf[slide_id])
if args.acc: