import numpy as np
import os
import json
import matplotlib.pyplot as plt
if __name__ == '__main__':
nepisodes = 500
lambda_param = 1
return_type = 'Retrace'
datadir = 'data'
if not os.path.exists(datadir):
os.mkdir(datadir)
print('collecting data ...')
collect_dataset(datadir)
print('estimating true Q function ...')
with open(os.path.join(datadir, 'test_points.json'), 'r') as f:
test_points = json.load(f)
estimate_true_q(test_points, datadir, nepisodes=100)
with open(os.path.join(datadir, 'test_points.json'), 'r') as f:
test_points = json.load(f)
data = np.load(os.path.join(datadir, 'dataset_0.npy'))
true_q = np.load(os.path.join(datadir, 'true_q.npy'))
value_function = ValueFunction()
print('estimating key quantities ...')
test_data = np.load(os.path.join(datadir, 'dataset_0.npy'))
A, b, M_inv = estimate_key_quantities(value_function, test_data,
lambda_param, return_type)
import cv2 from WebcamCapture import WebcamCapture from FaceDetection import FaceDetection from ProcessImage import ProcessImage from utils import collect_dataset images, labels, label_dic = collect_dataset() rec_lbph = cv2.face.LBPHFaceRecognizer_create() rec_lbph.train(images, labels) camera = WebcamCapture() face_detector = FaceDetection() image_processor = ProcessImage() images, labels, label_dic = collect_dataset() rec_lbph = cv2.face.LBPHFaceRecognizer_create() rec_lbph.train(images, labels) rec_eigen = cv2.face.EigenFaceRecognizer_create() rec_eigen.train(images, labels) rec_fisher = cv2.face.FisherFaceRecognizer_create() rec_fisher.train(images, labels) collector = cv2.face.StandardCollector_create() frame = camera.get_feed() faces_area = face_detector.detect_face(frame) if len(faces_area):
self.models[i][0]) + self.models[i][1]
prediction_result[inds] = np.argmax(predict_values, axis=1)
#print (prediction_result.shape)
#print (prediction_result)
return prediction_result
if __name__ == "__main__":
from actor_wrapper import Actor_LIME
from utils import collect_dataset
from sklearn.model_selection import train_test_split
actor = Actor_LIME(
nn_model="../pensieve_test/models/pretrain_linear_reward.ckpt")
dataset = collect_dataset().values
labels = np.argmax(actor.predict(dataset), axis=1)
X_train, X_test, y_train, y_test = train_test_split(dataset,
labels,
test_size=0.2)
print(X_train.shape, X_test.shape, y_train.shape, y_test.shape)
with open("./lime_extended_performance.csv", "w") as FILE:
for j in range(100):
X_train, X_test, y_train, y_test = train_test_split(dataset,
labels,
test_size=0.2)
for i in range(1, 50):
lime_model = LIMESimpleModel(cluster_num=i)