Skip to content

Instantly share code, notes, and snippets.

@KyoungHa-Park

KyoungHa-Park/gist:f9f6492195c9044a0a88ba415464207d

Created March 17, 2018 09:33
Show Gist options
  • Save KyoungHa-Park/f9f6492195c9044a0a88ba415464207d to your computer and use it in GitHub Desktop.
Save KyoungHa-Park/f9f6492195c9044a0a88ba415464207d to your computer and use it in GitHub Desktop.
Download ZIP
Raw
gistfile1.txt
{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"180315_dropout.ipynb","version":"0.3.2","views":{},"default_view":{},"provenance":[]}},"cells":[{"metadata":{"id":"BeR5YKCQOS2i","colab_type":"text"},"cell_type":"markdown","source":["Dropout\n","![대체 텍스트](https://i.imgur.com/zmENnkn.png)"]},{"metadata":{"id":"XLmTBOOM6H_W","colab_type":"text"},"cell_type":"markdown","source":["Setting up the data and the model\n","+ Data Preprocessing\n","+ Weight Initialization\n","+ Batch Normalization\n","+ Regularization (L2/L1/Maxnorm/Dropout)\n","\n","Loss functions\n","\n","Summary"]},{"metadata":{"id":"49tblNlP8qJJ","colab_type":"text"},"cell_type":"markdown","source":["##1.Dropout을 이용하는 이유는 Overfitting을 막기 위함이다.\n","\n"," + Dropout과 Overfitting의 특징은?\n"," + Overfitting이 정확히 언제(시기 혹은 과정) 보이는가?\n"," + Overfitting이 발생하는 원인은 무엇인가?\n"," + Overfitting은 왜 막아야 하는가?\n","---\n"]},{"metadata":{"id":"bEct7yrW7feO","colab_type":"text"},"cell_type":"markdown","source":["+ Dropout과 Overfitting의 특징은?\n"," + Overfitting은 신경망이 학습 데이터를 보고 패턴을 인식하는 것이 아니라 그냥 데이터를 외워버리면서 발생하는 문제이다.\n"," + 일부 정보를 누락하는 Dropout을 통해, 암기에서 학습으로 가는 방법(아이디어)임.\n"," + 수리적으로는 패널티 부과를 통해, 해당 값(weight)을 0으로 만든다\n"," > ![대체 텍스트](https://i.imgur.com/ltXEPYk.png)\n","\n","+ Overfitting이 정확히 언제(시기 혹은 과정) 보이는가?\n"," + 학습하는 과정에서 발생한다.\n"," + hidden visible layer를 가리지 않고 적용된다.\n","\n","+ Overfitting이 발생하는 원인은 무엇인가?\n"," + 수리적으로는 행렬곱 과정에서 변수간 연관성이 있는 경우, 역행렬로 만드는 과정에서 극단값으로 계산된다. 이를 “분사이 커진다”라고 표현하기도 하고,”변동성이 커진다”고 표현하기도 한다.\n"," + 통계적으로는 분산이 증가하면서, 모형의 예측력이 떨어진다고 보기 때문이다.\n","\n","+ Overfitting은 왜 막아야 하는가?\n"," + **예측**을 위해서는 암기가 아닌 **학습**이 필요하다.\n"," + 그래서 학습을 위해 일부 connect(논리간 연결고리)를 제외하므로서, 학습을 강제로 유도할 필요가 있다.\n"," + 혹은 학습의 일부 내용을 drop(제거 혹은 패널티)한다.\n"," + 결국, 학습 과정에서는 반드시 틀리는 경우가 발생한다.\n","\n","따라서 **Dropout**는 학습 과정에서 node 생략 등을 진행, **Coss function에 penalty 부과** 하는 방법중 하나이며, 이를 통해 예측이 가능한 모델(=more robust features)로 만드는 알고리즘이다."]},{"metadata":{"id":"Tpk8-sVv6s6_","colab_type":"text"},"cell_type":"markdown","source":["\n","##2.Overfitting을 막는 방법(들)을 Regulization이라 한다\n"," + Regulization의 원리(아이디어)는 정확히 무엇인가?\n"," + Regulization의 종류는 dropout뿐인가?\n"," + Regulization은 Normalization과 어떻게 다른가?\n"," + Regulization은 Generalization과 어떻게 다른가?\n","---\n","\n","+ Regulization의 원리(아이디어)는 정확히 무엇인가?\n"," + 회귀분석의 다중공선성문제 해법에서 나왔다\n"," + 다중공선성이란, 변수간 상관관계가 커서, 이를 통한 회귀분석의 결과가 신뢰하기 어려운(= 추정치의 분산이 크다) 현상을 뜻한다.\n"," + 다중공선성을 해결하기 위해서는 1) 상관성이 높은 변수 일부를 제거하는 방법을 우선시 한다. 그러나, 해당 변수가 중요한 변수이기 때문에 제거하지 못하는 경우, 2) 차선책으로서 편향추정방법인 능형회귀를 이용한다. 3)근레에는 능형회귀보다 좋다는 LASSO를 이용한다.\n"]},{"metadata":{"id":"ybJs_0bQ-Vtd","colab_type":"text"},"cell_type":"markdown","source":["+ Regulization의 종류는 dropout뿐인가?\n"," + 접근 방법이 다른 Case로는 L1 Regulization, L2 Regulization, Affine transformation과 Early stopping이 있다.\n"," + 이중에서 통계학적으로 접근한 것이 L1 Regulization과 L2 Regulization이며, 데이터 측면에서 접근한 Affine transformation과 Early stopping이 있다. \n"," > ![대체 텍스트](https://i.imgur.com/Ms1BQuv.jpg)\n"," + Dropout 세부적으로는 (학습시) node를 제외하는 dropout와 (학습시) 계산을 제외하는 dropconnect가 있다.\n"," > ![대체 텍스트](https://i.imgur.com/zmENnkn.png)\n"," + 각 drop 적용시, 확률분포에 대한 선택문제는 추가로 발생한다.\n"," > ![대체 텍스트](https://i.imgur.com/spRGQHv.png)\n"]},{"metadata":{"id":"M8NVDylr-X1M","colab_type":"text"},"cell_type":"markdown","source":["+ Regulization은 Normalization과 어떻게 다른가?\n"," + Gaussain DropoutConnect의 경우, dropout 할지 말지를 Nrom dist를 통해 구할 뿐이다.\n"," + Normalization은 위 결과 전/후에 대해 Sacle을 맞추어 주는 방법이다.\n"," + 따라서, 현재로써는 Regulization과 Normalization은 연결지어 보기는 어렵다.\n"," + 다만, 전문적인 영역에서는 구분짓는 상황이므로, 이부분에 대해서는 추가 확인 필요\n"," (https://www.semanticscholar.org/paper/L2-Regularization-versus-Batch-and-Weight-Normaliz-Laarhoven/7379455504274d7ff46a37bb4a0ccf81b60df86f)"]},{"metadata":{"id":"tUN8uXsq-X5w","colab_type":"text"},"cell_type":"markdown","source":["+ Regulization은 Generalization과 어떻게 다른가?\n"," + Generalization을 미시적으로 본 것이 Regulization이다.\n"," + Regulization까지 적용된 여러 모델들이 앙상블기법 이용해 General한 모델이 되는 것이다.\n"]},{"metadata":{"id":"7EVAgRX26J5c","colab_type":"code","colab":{"autoexec":{"startup":false,"wait_interval":0}}},"cell_type":"code","source":["# 06 - MNIST/ 02 - Dropout.py 코드 사용\n","# 과적합 방지 기법 중 하나인 Dropout 을 사용해봅니다.\n","# 우선은 변수설정에 대한 실행\n","\n","import tensorflow as tf\n","import numpy as np\n","import matplotlib.pyplot as plt\n","from datetime import datetime, timedelta\n","\n","\n","from tensorflow.examples.tutorials.mnist import input_data\n","mnist = input_data.read_data_sets(\"./mnist/data/\", one_hot=True)\n","\n","#########\n","# 신경망 모델 구성\n","######\n","X = tf.placeholder(tf.float32, [None, 784])\n","Y = tf.placeholder(tf.float32, [None, 10])\n","keep_prob = tf.placeholder(tf.float32)\n","\n","W1 = tf.Variable(tf.random_normal([784, 256], stddev=0.01))\n","L1 = tf.nn.relu(tf.matmul(X, W1))\n","# 텐서플로우에 내장된 함수를 이용하여 dropout 을 적용합니다.\n","# 함수에 적용할 레이어와 확률만 넣어주면 됩니다. 겁나 매직!!\n","L1 = tf.nn.dropout(L1, keep_prob)\n","\n","W2 = tf.Variable(tf.random_normal([256, 256], stddev=0.01))\n","L2 = tf.nn.relu(tf.matmul(L1, W2))\n","L2 = tf.nn.dropout(L2, keep_prob)\n","\n","W3 = tf.Variable(tf.random_normal([256, 10], stddev=0.01))\n","model = tf.matmul(L2, W3)\n","\n","cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=model, labels=Y))\n","optimizer = tf.train.AdamOptimizer(0.001).minimize(cost)\n"],"execution_count":0,"outputs":[]},{"metadata":{"id":"6a2IBWR7bwl4","colab_type":"code","colab":{"autoexec":{"startup":false,"wait_interval":0},"output_extras":[{"item_id":49},{"item_id":97},{"item_id":146},{"item_id":151}],"base_uri":"https://localhost:8080/","height":3108},"outputId":"15d8090f-12d1-4785-8283-6048ad4a5600","executionInfo":{"status":"ok","timestamp":1521270373238,"user_tz":-540,"elapsed":383273,"user":{"displayName":"박경하","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s128","userId":"110612470497084106049"}}},"cell_type":"code","source":["#########\n","# 신경망 모델0 학습 : dropout rate : 0.01\n","######\n","init = tf.global_variables_initializer()\n","sess0 = tf.Session()\n","sess0.run(init)\n","\n","batch_size = 100\n","total_batch = int(mnist.train.num_examples / batch_size)\n","\n","print('======================================')\n","print('모델0 최적화 시작!(dropout rate = 0.01)')\n","print('======================================')\n","\n","for epoch in range(30):\n"," total_cost = 0\n","\n"," for i in range(total_batch):\n"," batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n","\n"," _, cost_val = sess0.run([optimizer, cost],feed_dict={X: batch_xs,Y: batch_ys,keep_prob: 0.01}) # dropout 설정\n"," total_cost += cost_val\n","\n"," now = datetime.today()+ timedelta(hours =9) # 현재 in google\n","\n"," print('Epoch:', '%04d' % (epoch + 1),\n"," 'Avg. cost =', '{:.4f}'.format(total_cost / total_batch), 'Time =', now.strftime('%Y-%m-%d') +\" \" + now.strftime('%H:%M:%S'))\n"," \n","print('모델0 최적화 완료!(dropout rate = 0.01)')\n","\n","\n","\n","\n","#########\n","# 신경망 모델1 학습 : dropout rate : 0.2\n","######\n","init = tf.global_variables_initializer()\n","sess1 = tf.Session()\n","sess1.run(init)\n","\n","batch_size = 100\n","total_batch = int(mnist.train.num_examples / batch_size)\n","\n","print('======================================')\n","print('모델1 최적화 시작!(dropout rate = 0.2)')\n","print('======================================')\n","\n","for epoch in range(30):\n"," total_cost = 0\n","\n"," for i in range(total_batch):\n"," batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n","\n"," _, cost_val = sess1.run([optimizer, cost],feed_dict={X: batch_xs,Y: batch_ys,keep_prob: 0.2}) # dropout 설정\n"," total_cost += cost_val\n","\n"," now = datetime.today()+ timedelta(hours =9) # 현재 in google\n","\n"," print('Epoch:', '%04d' % (epoch + 1),\n"," 'Avg. cost =', '{:.4f}'.format(total_cost / total_batch), 'Time =', now.strftime('%Y-%m-%d') +\" \" + now.strftime('%H:%M:%S'))\n"," \n","print('모델1 최적화 완료!(dropout rate = 0.2)')\n","\n","\n","#########\n","# 신경망 모델2 학습 : dropout rate : 0.5\n","######\n","init = tf.global_variables_initializer()\n","sess2 = tf.Session()\n","sess2.run(init)\n","\n","batch_size = 100\n","total_batch = int(mnist.train.num_examples / batch_size)\n","\n","print('======================================')\n","print('모델2 최적화 시작!(dropout rate = 0.5)')\n","print('======================================')\n","\n","for epoch in range(30):\n"," total_cost = 0\n","\n"," for i in range(total_batch):\n"," batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n","\n"," _, cost_val = sess2.run([optimizer, cost],feed_dict={X: batch_xs,Y: batch_ys,keep_prob: 0.5}) # dropout 설정\n"," total_cost += cost_val\n","\n"," now = datetime.today()+ timedelta(hours =9) # 현재 in google\n","\n"," print('Epoch:', '%04d' % (epoch + 1),\n"," 'Avg. cost =', '{:.4f}'.format(total_cost / total_batch), 'Time =', now.strftime('%Y-%m-%d') +\" \" + now.strftime('%H:%M:%S'))\n"," \n","print('모델2 최적화 완료!(dropout rate = 0.5)')\n","\n","\n","\n","\n","#########\n","# 신경망 모델3 학습 : dropout rate : 0.8\n","######\n","init = tf.global_variables_initializer()\n","sess3 = tf.Session()\n","sess3.run(init)\n","\n","batch_size = 100\n","total_batch = int(mnist.train.num_examples / batch_size)\n","\n","print('======================================')\n","print('모델3 최적화 시작!(dropout rate = 0.8)')\n","print('======================================')\n","for epoch in range(30):\n"," total_cost = 0\n","\n"," for i in range(total_batch):\n"," batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n","\n"," _, cost_val = sess3.run([optimizer, cost],feed_dict={X: batch_xs,Y: batch_ys,keep_prob: 0.8}) # dropout 설정\n","\n"," total_cost += cost_val\n","\n"," now = datetime.today()+ timedelta(hours =9) # 현재 in google\n"," date = now.strftime('%Y-%m-%d')\n"," time = now.strftime('%H:%M:%S')\n"," \n"," print('Epoch:', '%04d' % (epoch + 1),\n"," 'Avg. cost =', '{:.4f}'.format(total_cost / total_batch), 'Time =', now.strftime('%Y-%m-%d') +\" \" + now.strftime('%H:%M:%S'))\n"," \n","print('모델3 최적화 완료!(dropout rate = 0.8)')\n","\n","\n","\n","\n","\n","#########\n","# 신경망 모델4 학습 : dropout rate : 0.99\n","######\n","init = tf.global_variables_initializer()\n","sess4 = tf.Session()\n","sess4.run(init)\n","\n","batch_size = 100\n","total_batch = int(mnist.train.num_examples / batch_size)\n","\n","print('======================================')\n","print('모델4 최적화 시작!(dropout rate = 0.99)')\n","print('======================================')\n","for epoch in range(30):\n"," total_cost = 0\n","\n"," for i in range(total_batch):\n"," batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n","\n"," _, cost_val = sess4.run([optimizer, cost],feed_dict={X: batch_xs,Y: batch_ys,keep_prob: 0.99}) # dropout 설정\n","\n"," total_cost += cost_val\n","\n"," now = datetime.today()+ timedelta(hours =9) # 현재 in google\n"," date = now.strftime('%Y-%m-%d')\n"," time = now.strftime('%H:%M:%S')\n"," \n"," print('Epoch:', '%04d' % (epoch + 1),\n"," 'Avg. cost =', '{:.4f}'.format(total_cost / total_batch), 'Time =', now.strftime('%Y-%m-%d') +\" \" + now.strftime('%H:%M:%S'))\n"," \n","print('모델3 최적화 완료!(dropout rate = 0.99)')\n","\n"],"execution_count":28,"outputs":[{"output_type":"stream","text":["======================================\n","모델0 최적화 시작!(dropout rate = 0.01)\n","======================================\n","('Epoch:', '0001', 'Avg. cost =', '2.3082', 'Time =', '2018-03-17 15:59:51')\n","('Epoch:', '0002', 'Avg. cost =', '2.3033', 'Time =', '2018-03-17 15:59:53')\n","('Epoch:', '0003', 'Avg. cost =', '2.3037', 'Time =', '2018-03-17 15:59:56')\n","('Epoch:', '0004', 'Avg. cost =', '2.3030', 'Time =', '2018-03-17 15:59:58')\n","('Epoch:', '0005', 'Avg. cost =', '2.3035', 'Time =', '2018-03-17 16:00:01')\n","('Epoch:', '0006', 'Avg. cost =', '2.3031', 'Time =', '2018-03-17 16:00:04')\n","('Epoch:', '0007', 'Avg. cost =', '2.3029', 'Time =', '2018-03-17 16:00:06')\n","('Epoch:', '0008', 'Avg. cost =', '2.3029', 'Time =', '2018-03-17 16:00:09')\n","('Epoch:', '0009', 'Avg. cost =', '2.3028', 'Time =', '2018-03-17 16:00:11')\n","('Epoch:', '0010', 'Avg. cost =', '2.3025', 'Time =', '2018-03-17 16:00:14')\n","('Epoch:', '0011', 'Avg. cost =', '2.3028', 'Time =', '2018-03-17 16:00:17')\n","('Epoch:', '0012', 'Avg. cost =', '2.3025', 'Time =', '2018-03-17 16:00:19')\n","('Epoch:', '0013', 'Avg. cost =', '2.3026', 'Time =', '2018-03-17 16:00:22')\n","('Epoch:', '0014', 'Avg. cost =', '2.3038', 'Time =', '2018-03-17 16:00:24')\n","('Epoch:', '0015', 'Avg. cost =', '2.3027', 'Time =', '2018-03-17 16:00:27')\n","('Epoch:', '0016', 'Avg. cost =', '2.3025', 'Time =', '2018-03-17 16:00:30')\n","('Epoch:', '0017', 'Avg. cost =', '2.3023', 'Time =', '2018-03-17 16:00:32')\n","('Epoch:', '0018', 'Avg. cost =', '2.3027', 'Time =', '2018-03-17 16:00:35')\n","('Epoch:', '0019', 'Avg. cost =', '2.3033', 'Time =', '2018-03-17 16:00:37')\n","('Epoch:', '0020', 'Avg. cost =', '2.3022', 'Time =', '2018-03-17 16:00:40')\n","('Epoch:', '0021', 'Avg. cost =', '2.3021', 'Time =', '2018-03-17 16:00:42')\n","('Epoch:', '0022', 'Avg. cost =', '2.3025', 'Time =', '2018-03-17 16:00:45')\n","('Epoch:', '0023', 'Avg. cost =', '2.3019', 'Time =', '2018-03-17 16:00:47')\n","('Epoch:', '0024', 'Avg. cost =', '2.3022', 'Time =', '2018-03-17 16:00:50')\n","('Epoch:', '0025', 'Avg. cost =', '2.3026', 'Time =', '2018-03-17 16:00:52')\n","('Epoch:', '0026', 'Avg. cost =', '2.3022', 'Time =', '2018-03-17 16:00:55')\n","('Epoch:', '0027', 'Avg. cost =', '2.3017', 'Time =', '2018-03-17 16:00:57')\n","('Epoch:', '0028', 'Avg. cost =', '2.3024', 'Time =', '2018-03-17 16:01:00')\n","('Epoch:', '0029', 'Avg. cost =', '2.3021', 'Time =', '2018-03-17 16:01:02')\n","('Epoch:', '0030', 'Avg. cost =', '2.3021', 'Time =', '2018-03-17 16:01:05')\n","모델0 최적화 완료!(dropout rate = 0.01)\n","======================================\n","모델1 최적화 시작!(dropout rate = 0.2)\n","======================================\n","('Epoch:', '0001', 'Avg. cost =', '0.8583', 'Time =', '2018-03-17 16:01:08')\n","('Epoch:', '0002', 'Avg. cost =', '0.5032', 'Time =', '2018-03-17 16:01:10')\n","('Epoch:', '0003', 'Avg. cost =', '0.4510', 'Time =', '2018-03-17 16:01:13')\n","('Epoch:', '0004', 'Avg. cost =', '0.4155', 'Time =', '2018-03-17 16:01:15')\n","('Epoch:', '0005', 'Avg. cost =', '0.4012', 'Time =', '2018-03-17 16:01:18')\n","('Epoch:', '0006', 'Avg. cost =', '0.3886', 'Time =', '2018-03-17 16:01:20')\n","('Epoch:', '0007', 'Avg. cost =', '0.3761', 'Time =', '2018-03-17 16:01:23')\n","('Epoch:', '0008', 'Avg. cost =', '0.3651', 'Time =', '2018-03-17 16:01:26')\n","('Epoch:', '0009', 'Avg. cost =', '0.3589', 'Time =', '2018-03-17 16:01:28')\n","('Epoch:', '0010', 'Avg. cost =', '0.3493', 'Time =', '2018-03-17 16:01:31')\n","('Epoch:', '0011', 'Avg. cost =', '0.3441', 'Time =', '2018-03-17 16:01:33')\n","('Epoch:', '0012', 'Avg. cost =', '0.3359', 'Time =', '2018-03-17 16:01:36')\n","('Epoch:', '0013', 'Avg. cost =', '0.3367', 'Time =', '2018-03-17 16:01:38')\n","('Epoch:', '0014', 'Avg. cost =', '0.3326', 'Time =', '2018-03-17 16:01:41')\n","('Epoch:', '0015', 'Avg. cost =', '0.3261', 'Time =', '2018-03-17 16:01:43')\n","('Epoch:', '0016', 'Avg. cost =', '0.3294', 'Time =', '2018-03-17 16:01:46')\n","('Epoch:', '0017', 'Avg. cost =', '0.3266', 'Time =', '2018-03-17 16:01:48')\n","('Epoch:', '0018', 'Avg. cost =', '0.3221', 'Time =', '2018-03-17 16:01:51')\n"],"name":"stdout"},{"output_type":"stream","text":["('Epoch:', '0019', 'Avg. cost =', '0.3231', 'Time =', '2018-03-17 16:01:53')\n","('Epoch:', '0020', 'Avg. cost =', '0.3130', 'Time =', '2018-03-17 16:01:56')\n","('Epoch:', '0021', 'Avg. cost =', '0.3115', 'Time =', '2018-03-17 16:01:58')\n","('Epoch:', '0022', 'Avg. cost =', '0.3079', 'Time =', '2018-03-17 16:02:01')\n","('Epoch:', '0023', 'Avg. cost =', '0.3096', 'Time =', '2018-03-17 16:02:03')\n","('Epoch:', '0024', 'Avg. cost =', '0.3072', 'Time =', '2018-03-17 16:02:06')\n","('Epoch:', '0025', 'Avg. cost =', '0.3065', 'Time =', '2018-03-17 16:02:08')\n","('Epoch:', '0026', 'Avg. cost =', '0.3039', 'Time =', '2018-03-17 16:02:11')\n","('Epoch:', '0027', 'Avg. cost =', '0.3020', 'Time =', '2018-03-17 16:02:14')\n","('Epoch:', '0028', 'Avg. cost =', '0.3043', 'Time =', '2018-03-17 16:02:16')\n","('Epoch:', '0029', 'Avg. cost =', '0.2960', 'Time =', '2018-03-17 16:02:19')\n","('Epoch:', '0030', 'Avg. cost =', '0.2944', 'Time =', '2018-03-17 16:02:21')\n","모델1 최적화 완료!(dropout rate = 0.2)\n","======================================\n","모델2 최적화 시작!(dropout rate = 0.5)\n","======================================\n","('Epoch:', '0001', 'Avg. cost =', '0.5187', 'Time =', '2018-03-17 16:02:24')\n","('Epoch:', '0002', 'Avg. cost =', '0.2341', 'Time =', '2018-03-17 16:02:26')\n","('Epoch:', '0003', 'Avg. cost =', '0.1831', 'Time =', '2018-03-17 16:02:29')\n","('Epoch:', '0004', 'Avg. cost =', '0.1602', 'Time =', '2018-03-17 16:02:31')\n","('Epoch:', '0005', 'Avg. cost =', '0.1406', 'Time =', '2018-03-17 16:02:34')\n","('Epoch:', '0006', 'Avg. cost =', '0.1273', 'Time =', '2018-03-17 16:02:37')\n","('Epoch:', '0007', 'Avg. cost =', '0.1207', 'Time =', '2018-03-17 16:02:39')\n","('Epoch:', '0008', 'Avg. cost =', '0.1118', 'Time =', '2018-03-17 16:02:42')\n","('Epoch:', '0009', 'Avg. cost =', '0.1053', 'Time =', '2018-03-17 16:02:44')\n","('Epoch:', '0010', 'Avg. cost =', '0.1029', 'Time =', '2018-03-17 16:02:47')\n","('Epoch:', '0011', 'Avg. cost =', '0.0976', 'Time =', '2018-03-17 16:02:49')\n","('Epoch:', '0012', 'Avg. cost =', '0.0930', 'Time =', '2018-03-17 16:02:52')\n","('Epoch:', '0013', 'Avg. cost =', '0.0881', 'Time =', '2018-03-17 16:02:54')\n","('Epoch:', '0014', 'Avg. cost =', '0.0864', 'Time =', '2018-03-17 16:02:57')\n","('Epoch:', '0015', 'Avg. cost =', '0.0827', 'Time =', '2018-03-17 16:03:00')\n","('Epoch:', '0016', 'Avg. cost =', '0.0838', 'Time =', '2018-03-17 16:03:02')\n","('Epoch:', '0017', 'Avg. cost =', '0.0797', 'Time =', '2018-03-17 16:03:05')\n","('Epoch:', '0018', 'Avg. cost =', '0.0795', 'Time =', '2018-03-17 16:03:07')\n","('Epoch:', '0019', 'Avg. cost =', '0.0783', 'Time =', '2018-03-17 16:03:10')\n","('Epoch:', '0020', 'Avg. cost =', '0.0741', 'Time =', '2018-03-17 16:03:12')\n","('Epoch:', '0021', 'Avg. cost =', '0.0777', 'Time =', '2018-03-17 16:03:15')\n","('Epoch:', '0022', 'Avg. cost =', '0.0731', 'Time =', '2018-03-17 16:03:17')\n","('Epoch:', '0023', 'Avg. cost =', '0.0720', 'Time =', '2018-03-17 16:03:20')\n","('Epoch:', '0024', 'Avg. cost =', '0.0709', 'Time =', '2018-03-17 16:03:22')\n","('Epoch:', '0025', 'Avg. cost =', '0.0689', 'Time =', '2018-03-17 16:03:25')\n","('Epoch:', '0026', 'Avg. cost =', '0.0640', 'Time =', '2018-03-17 16:03:28')\n","('Epoch:', '0027', 'Avg. cost =', '0.0641', 'Time =', '2018-03-17 16:03:30')\n","('Epoch:', '0028', 'Avg. cost =', '0.0717', 'Time =', '2018-03-17 16:03:33')\n","('Epoch:', '0029', 'Avg. cost =', '0.0635', 'Time =', '2018-03-17 16:03:35')\n","('Epoch:', '0030', 'Avg. cost =', '0.0635', 'Time =', '2018-03-17 16:03:38')\n","모델2 최적화 완료!(dropout rate = 0.5)\n","======================================\n","모델3 최적화 시작!(dropout rate = 0.8)\n","======================================\n","('Epoch:', '0001', 'Avg. cost =', '0.4326', 'Time =', '2018-03-17 16:03:40')\n","('Epoch:', '0002', 'Avg. cost =', '0.1662', 'Time =', '2018-03-17 16:03:43')\n","('Epoch:', '0003', 'Avg. cost =', '0.1159', 'Time =', '2018-03-17 16:03:46')\n","('Epoch:', '0004', 'Avg. cost =', '0.0882', 'Time =', '2018-03-17 16:03:48')\n","('Epoch:', '0005', 'Avg. cost =', '0.0730', 'Time =', '2018-03-17 16:03:51')\n","('Epoch:', '0006', 'Avg. cost =', '0.0609', 'Time =', '2018-03-17 16:03:53')\n"],"name":"stdout"},{"output_type":"stream","text":["('Epoch:', '0007', 'Avg. cost =', '0.0511', 'Time =', '2018-03-17 16:03:56')\n","('Epoch:', '0008', 'Avg. cost =', '0.0450', 'Time =', '2018-03-17 16:03:58')\n","('Epoch:', '0009', 'Avg. cost =', '0.0421', 'Time =', '2018-03-17 16:04:01')\n","('Epoch:', '0010', 'Avg. cost =', '0.0367', 'Time =', '2018-03-17 16:04:04')\n","('Epoch:', '0011', 'Avg. cost =', '0.0332', 'Time =', '2018-03-17 16:04:06')\n","('Epoch:', '0012', 'Avg. cost =', '0.0300', 'Time =', '2018-03-17 16:04:09')\n","('Epoch:', '0013', 'Avg. cost =', '0.0295', 'Time =', '2018-03-17 16:04:11')\n","('Epoch:', '0014', 'Avg. cost =', '0.0276', 'Time =', '2018-03-17 16:04:14')\n","('Epoch:', '0015', 'Avg. cost =', '0.0243', 'Time =', '2018-03-17 16:04:16')\n","('Epoch:', '0016', 'Avg. cost =', '0.0247', 'Time =', '2018-03-17 16:04:19')\n","('Epoch:', '0017', 'Avg. cost =', '0.0221', 'Time =', '2018-03-17 16:04:21')\n","('Epoch:', '0018', 'Avg. cost =', '0.0215', 'Time =', '2018-03-17 16:04:24')\n","('Epoch:', '0019', 'Avg. cost =', '0.0214', 'Time =', '2018-03-17 16:04:26')\n","('Epoch:', '0020', 'Avg. cost =', '0.0197', 'Time =', '2018-03-17 16:04:29')\n","('Epoch:', '0021', 'Avg. cost =', '0.0188', 'Time =', '2018-03-17 16:04:31')\n","('Epoch:', '0022', 'Avg. cost =', '0.0195', 'Time =', '2018-03-17 16:04:34')\n","('Epoch:', '0023', 'Avg. cost =', '0.0187', 'Time =', '2018-03-17 16:04:37')\n","('Epoch:', '0024', 'Avg. cost =', '0.0193', 'Time =', '2018-03-17 16:04:39')\n","('Epoch:', '0025', 'Avg. cost =', '0.0177', 'Time =', '2018-03-17 16:04:42')\n","('Epoch:', '0026', 'Avg. cost =', '0.0151', 'Time =', '2018-03-17 16:04:44')\n","('Epoch:', '0027', 'Avg. cost =', '0.0161', 'Time =', '2018-03-17 16:04:47')\n","('Epoch:', '0028', 'Avg. cost =', '0.0155', 'Time =', '2018-03-17 16:04:49')\n","('Epoch:', '0029', 'Avg. cost =', '0.0172', 'Time =', '2018-03-17 16:04:52')\n","('Epoch:', '0030', 'Avg. cost =', '0.0143', 'Time =', '2018-03-17 16:04:55')\n","모델3 최적화 완료!(dropout rate = 0.8)\n","======================================\n","모델4 최적화 시작!(dropout rate = 0.99)\n","======================================\n","('Epoch:', '0001', 'Avg. cost =', '0.4023', 'Time =', '2018-03-17 16:04:57')\n","('Epoch:', '0002', 'Avg. cost =', '0.1467', 'Time =', '2018-03-17 16:05:00')\n","('Epoch:', '0003', 'Avg. cost =', '0.0948', 'Time =', '2018-03-17 16:05:02')\n","('Epoch:', '0004', 'Avg. cost =', '0.0705', 'Time =', '2018-03-17 16:05:05')\n","('Epoch:', '0005', 'Avg. cost =', '0.0526', 'Time =', '2018-03-17 16:05:07')\n","('Epoch:', '0006', 'Avg. cost =', '0.0393', 'Time =', '2018-03-17 16:05:10')\n","('Epoch:', '0007', 'Avg. cost =', '0.0321', 'Time =', '2018-03-17 16:05:12')\n","('Epoch:', '0008', 'Avg. cost =', '0.0259', 'Time =', '2018-03-17 16:05:15')\n","('Epoch:', '0009', 'Avg. cost =', '0.0226', 'Time =', '2018-03-17 16:05:17')\n","('Epoch:', '0010', 'Avg. cost =', '0.0190', 'Time =', '2018-03-17 16:05:20')\n","('Epoch:', '0011', 'Avg. cost =', '0.0179', 'Time =', '2018-03-17 16:05:23')\n","('Epoch:', '0012', 'Avg. cost =', '0.0152', 'Time =', '2018-03-17 16:05:25')\n","('Epoch:', '0013', 'Avg. cost =', '0.0099', 'Time =', '2018-03-17 16:05:28')\n","('Epoch:', '0014', 'Avg. cost =', '0.0130', 'Time =', '2018-03-17 16:05:30')\n","('Epoch:', '0015', 'Avg. cost =', '0.0141', 'Time =', '2018-03-17 16:05:33')\n","('Epoch:', '0016', 'Avg. cost =', '0.0091', 'Time =', '2018-03-17 16:05:35')\n","('Epoch:', '0017', 'Avg. cost =', '0.0101', 'Time =', '2018-03-17 16:05:38')\n","('Epoch:', '0018', 'Avg. cost =', '0.0101', 'Time =', '2018-03-17 16:05:40')\n","('Epoch:', '0019', 'Avg. cost =', '0.0115', 'Time =', '2018-03-17 16:05:43')\n","('Epoch:', '0020', 'Avg. cost =', '0.0092', 'Time =', '2018-03-17 16:05:45')\n","('Epoch:', '0021', 'Avg. cost =', '0.0079', 'Time =', '2018-03-17 16:05:48')\n","('Epoch:', '0022', 'Avg. cost =', '0.0062', 'Time =', '2018-03-17 16:05:50')\n","('Epoch:', '0023', 'Avg. cost =', '0.0088', 'Time =', '2018-03-17 16:05:53')\n","('Epoch:', '0024', 'Avg. cost =', '0.0105', 'Time =', '2018-03-17 16:05:56')\n","('Epoch:', '0025', 'Avg. cost =', '0.0084', 'Time =', '2018-03-17 16:05:58')\n"],"name":"stdout"},{"output_type":"stream","text":["('Epoch:', '0026', 'Avg. cost =', '0.0054', 'Time =', '2018-03-17 16:06:01')\n","('Epoch:', '0027', 'Avg. cost =', '0.0065', 'Time =', '2018-03-17 16:06:03')\n","('Epoch:', '0028', 'Avg. cost =', '0.0100', 'Time =', '2018-03-17 16:06:06')\n","('Epoch:', '0029', 'Avg. cost =', '0.0054', 'Time =', '2018-03-17 16:06:08')\n","('Epoch:', '0030', 'Avg. cost =', '0.0024', 'Time =', '2018-03-17 16:06:11')\n","모델3 최적화 완료!(dropout rate = 0.99)\n"],"name":"stdout"}]},{"metadata":{"id":"oPm8Po-qbwoi","colab_type":"code","colab":{"autoexec":{"startup":false,"wait_interval":0},"output_extras":[{"item_id":1},{"item_id":2},{"item_id":3},{"item_id":4},{"item_id":5},{"item_id":6},{"item_id":7},{"item_id":8},{"item_id":9},{"item_id":10},{"item_id":11}],"base_uri":"https://localhost:8080/","height":1767},"outputId":"9bcb19a4-61be-488e-8b33-ec979c834e28","executionInfo":{"status":"ok","timestamp":1521271246428,"user_tz":-540,"elapsed":25800,"user":{"displayName":"박경하","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s128","userId":"110612470497084106049"}}},"cell_type":"code","source":["print('======================================================================')\n","#########\n","# 모델0 결과 확인\n","######\n","is_correct = tf.equal(tf.argmax(model, 1), tf.argmax(Y, 1))\n","accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))\n","print('Model1 (dropout rate = 0.01):', sess0.run(accuracy,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1}))\n","\n","#########\n","# 모델1 결과 확인 (matplot)\n","######\n","labels = sess0.run(model,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1})\n","\n","fig = plt.figure()\n","for i in range(100):\n"," subplot = fig.add_subplot(5, 20, i + 1)\n"," subplot.set_xticks([])\n"," subplot.set_yticks([])\n"," subplot.set_title('%d' % np.argmax(labels[i]))\n"," subplot.imshow(mnist.test.images[i].reshape((28, 28)),cmap=plt.cm.gray_r)\n","\n","plt.show()\n","print('======================================================================')\n","print(' ')\n","\n","\n","#########\n","# 모델1 결과 확인\n","######\n","is_correct = tf.equal(tf.argmax(model, 1), tf.argmax(Y, 1))\n","accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))\n","print('Model1 (dropout rate = 0.2):', sess1.run(accuracy,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1}))\n","\n","#########\n","# 모델1 결과 확인 (matplot)\n","######\n","labels = sess1.run(model,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1})\n","\n","fig = plt.figure()\n","for i in range(100):\n"," subplot = fig.add_subplot(5, 20, i + 1)\n"," subplot.set_xticks([])\n"," subplot.set_yticks([])\n"," subplot.set_title('%d' % np.argmax(labels[i]))\n"," subplot.imshow(mnist.test.images[i].reshape((28, 28)),cmap=plt.cm.gray_r)\n","\n","plt.show()\n","print('======================================================================')\n","print(' ')\n","\n","\n","#########\n","# 모델2 결과 확인\n","######\n","is_correct = tf.equal(tf.argmax(model, 1), tf.argmax(Y, 1))\n","accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))\n","print('Model2 (dropout rate = 0.5) :', sess2.run(accuracy,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1}))\n","\n","\n","#########\n","# 모델2 결과 확인 (matplot)\n","######\n","labels = sess2.run(model,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1})\n","\n","fig = plt.figure()\n","for i in range(100):\n"," subplot = fig.add_subplot(5, 20, i + 1)\n"," subplot.set_xticks([])\n"," subplot.set_yticks([])\n"," subplot.set_title('%d' % np.argmax(labels[i]))\n"," subplot.imshow(mnist.test.images[i].reshape((28, 28)),cmap=plt.cm.gray_r)\n","\n","plt.show()\n","print('======================================================================')\n","print(' ')\n","\n","\n","#########\n","# 모델3 결과 확인\n","######\n","is_correct = tf.equal(tf.argmax(model, 1), tf.argmax(Y, 1))\n","accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))\n","print('Model3 (dropout rate = 0.8)', sess3.run(accuracy,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1}))\n","\n","\n","#########\n","# 모델3 결과 확인 (matplot)\n","######\n","labels = sess3.run(model,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1})\n","\n","fig = plt.figure()\n","for i in range(100):\n"," subplot = fig.add_subplot(5, 20, i + 1)\n"," subplot.set_xticks([])\n"," subplot.set_yticks([])\n"," subplot.set_title('%d' % np.argmax(labels[i]))\n"," subplot.imshow(mnist.test.images[i].reshape((28, 28)),cmap=plt.cm.gray_r)\n","\n","plt.show()\n","print('======================================================================')\n","print(' ')\n","\n","\n","\n","#########\n","# 모델4 결과 확인\n","######\n","is_correct = tf.equal(tf.argmax(model, 1), tf.argmax(Y, 1))\n","accuracy = tf.reduce_mean(tf.cast(is_correct, tf.float32))\n","print('Model4 (dropout rate = 0.99):', sess4.run(accuracy,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1}))\n","\n","#########\n","# 모델1 결과 확인 (matplot)\n","######\n","labels = sess4.run(model,feed_dict={X: mnist.test.images,Y: mnist.test.labels,keep_prob: 1})\n","\n","fig = plt.figure()\n","for i in range(100):\n"," subplot = fig.add_subplot(5, 20, i + 1)\n"," subplot.set_xticks([])\n"," subplot.set_yticks([])\n"," subplot.set_title('%d' % np.argmax(labels[i]))\n"," subplot.imshow(mnist.test.images[i].reshape((28, 28)),cmap=plt.cm.gray_r)\n","\n","plt.show()\n","print('======================================================================')"],"execution_count":45,"outputs":[{"output_type":"stream","text":["======================================================================\n","('Model1 (dropout rate = 0.01):', 0.1573)\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAcwAAAEnCAYAAAAkWHPHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngofJr+8AAAAASUVORK5CYII=\n","text/plain":["<matplotlib.figure.Figure at 0x7f4a3f6e8a90>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["======================================================================\n"," \n","('Model1 (dropout rate = 0.2):', 0.9633)\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAcwAAAEnCAYAAAAkWHPHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n","text/plain":["<matplotlib.figure.Figure at 0x7f4a3e863490>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["======================================================================\n"," \n","('Model2 (dropout rate = 0.5) :', 0.9818)\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAcwAAAEnCAYAAAAkWHPHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n","text/plain":["<matplotlib.figure.Figure at 0x7f4a3db42c10>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["======================================================================\n"," \n","('Model3 (dropout rate = 0.8)', 0.9816)\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAcwAAAEnCAYAAAAkWHPHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n","text/plain":["<matplotlib.figure.Figure at 0x7f4a3dabb050>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["======================================================================\n"," \n","('Model4 (dropout rate = 0.99):', 0.9817)\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAcwAAAEnCAYAAAAkWHPHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzsvXl8Tdf+//9eW0gqQgwRmoTwiCQ+\nptCIpNEQVZKbkKY15RrjKuKi3+Sicmto66GlSou2hqorRYsqWqWmtMqVKKGDGhI0ISEDFZGJiLx+\nf/jtfTM75+y1Dfe+n4/Hfjxy9jnntd9Z+73Wa+211t5HAAAxDMMwDFMryqMOgGEYhmGeBNgwGYZh\nGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgQsNsw9e/ZQUFBQhc3Dw4MKCgp0BxUf\nH09hYWEUHBxMERERlJKSoltT5e7du7RgwQLy8PCgrKwsabqJiYkUHh5O/fv3p8jISKnaKgcPHiQP\nDw/KyMiQpmlUeajIjnnHjh0UEhJCvXv3punTp1NJSYkU3YyMDOrQoUOFfJ4xY4YUbaNiJjIu77Kz\nsykyMpL69OlDAwYMoOPHj0vRJTKuPIysg0bUEyNzjsi4cjay7d+7dy+FhYVRUFCQ9LafSEJ7BEns\n2rULkydP1q2TlZUFb29vnD9/HgCwYcMGDB06VLeuyrhx47B06VK4u7sjMzNTimZhYSF8fX3x+++/\nAwDi4uIwfvx4KdoqRUVFCA0NhY+PD9LT06XpGlEeKrJjTk5Oho+PD65evYqysjLExMTgww8/lBAp\nkJ6ejsDAQCla5TEyZiPzbsyYMVi7di0AIDExEVOnTpWia1R5GF0HjagnRuUcYGzeVUZW23/lyhX0\n6NEDGRkZAIB169bh5Zdf1q2rIqM9kjIke+fOHVq6dClNnz5dt5aVlRUtXryY3NzciIjomWeeoQsX\nLujWVZk0aRJNnTpVmh4R0dGjR8nFxYU6dOhAREQvv/wyHTlyREqPS2X58uU0cOBAsrW1laZJZEx5\nqMiO+ejRo+Tr60stW7YkIQSNHj2a9u3bJ0XbKIyM2ai8y8zMpNOnT9OIESOIiMjX15eWLl2qO14i\n48rD6DpoZD0xgodVV4xo+52cnIiIyM/Pj1JTU3Xrqshoj6QY5tatW6lbt27UqlUr3VpNmzalgIAA\n7fWhQ4eoS5cuunVVunbtKk1LJS0tjVxcXLTXtra2ZG9vT5cvX5ain5ycTAkJCTRmzBgpeuUxojyI\njIlZCEFlZWXa6/r160srYyKigoICmjRpEgUFBdHf/vY3unjxom5NI2M2Ku/OnTtHzs7OtHjxYurf\nvz+NGDGCzpw5ozdcIjKuPIyug0bVEyNyjsj4uqIis+1v3rw5+fv7ExFRaWkpbd++nZ5//nndukTy\n2iPdhllWVkZr166lsWPH6pWqQmJiIsXFxVFsbKx0bZkUFxeTtbV1hX3W1tZUVFSkWxsAzZ07l2bN\nmkV169bVrfcwMCpmPz8/OnLkCKWkpFBpaSlt3LiR7ty5I0Xb1taWQkND6Z///Cft3r2b/P39adKk\nSVRaWvrYxmxU3t26dYtSUlLI29ub9u7dSwMHDqTJkyfrLgsi48rDyDpoFEblHJGxeadiVNsfFxdH\n/v7+lJSURNOmTdOtJ7M90m2YP//8M9WvX5/atWunV6oCBw4coJkzZ9LKlSu14dnHlfr161dJxtu3\nb0sZity8eTO5ubmRt7e3bq2HhVExu7m50ezZsykmJoaGDBlCbm5uZGdnJ0W7cePGNGfOHHJ2diZF\nUSgyMpKuX79OaWlpj23MRuWdnZ0dNW3alPr27UtERIMHD6a8vDzdZUFkXHkYWQeNwqicIzI271SM\navtHjx5NR48epdGjR9OwYcPo9u3buvRktke6DfPgwYPUq1cv3YGUJyEhgebPn09r166lTp06SdU2\ngrZt21YY7sjPz6e8vDxq3bq1bu34+HiKj48nf39/8vf3p8zMTBo0aBAdPXpUt7ZRGBlzeHg4ffvt\nt7Rt2zZyd3cnd3d3CRET5eXlUXp6eoV9ZWVlZGVlpVvbqJiNyrunn36aCgsLtSE9IQQpikKKIucu\nNCPKw8g6aBRG5hyRcXmnIrvtv3jxIiUkJBDR/ZwLDQ2lwsJC3fOYUtsjvSuPxo0bh23btumV0Sgq\nKkJAQABOnTolTbM6ZK52Ky4uxrPPPovjx48DAJYtWyZl1Vh1BAYGSl0lq2LEKlkVWTGnpaVh4MCB\nyMvLQ0lJCcaOHSst9w4fPozAwED8+eefAIBNmzYhODgYpaWlunSNjNmovCsrK0NwcDA2bdoEANi9\nezf69OmDu3fv6tY2qjweVh2UWU+MyjnA2LxTkd32Hzt2DP7+/sjKygIAJCUloWvXrsjPz5d2DEBf\ne6S7K5OVlUXNmjXTK6MRHx9PN27cqDJ2vWHDBt3HuX79urbyj4ho5MiRVKdOHYqLiyNHR0eLdW1s\nbGjJkiX01ltvUXFxMbVq1YoWLFigK9aHgVHlYRStW7em559/nsLCwkgIQSEhIRQeHi5Fu2fPnvTX\nv/6VIiIiSAhBjo6OtHz5cqpTp85jG7NReSeEoGXLltHMmTNp9erV1LRpU1q6dKmUKx+jysPIOmhU\nPTEq54iMzTsV2W1/9+7dKSoqiiIjI6msrIzq1atH77//PjVo0EDaMfQiAP49TIZhGIZ5EPxoPIZh\nGIYxATZMhmEYhjEBNkyGYRiGMQE2TIZhGIYxATZMhmEYhjEBNkyGYRiGMYHH0jDnzZtHgwcPJiEE\nTZo0idavX/+oQ2KY/ymeeeYZaU/2YZj/Fh6rGjF06FBSFIXmzp1Lp0+fprCwMFq5ciWNHj3akONt\n3ryZtm3bJk3v/PnzVLduXVIUhT788EMpmq6urtXuN+Knenbu3Ekffvgh3bt3z2KNnJwc+uc//1nj\n8zDz8vJo586dFus/DL799luaMGGC9ji49u3b01NPPfVYGcitW7eofv361LRpUxJCaFv9+vXJwcFB\nl/aiRYuoadOm2mPKniRyc3NpxowZ2uP8ZsyYQdnZ2VK01TI+ePCgbq3Vq1dTTEwMde/eXcuzOnXq\nUFRUlP5AnxAKCgpo7ty52v8/cuRIunHjxkM7/rBhw2jDhg3mfUnqM4d0UrduXQghkJqaivz8fNy5\ncwdeXl4QQhhyvJkzZ0rTysnJgZ+fH6ysrCCEwPLly6XotmjRotr9f//736Xoq1y/fh1OTk4gIhQV\nFVmkcePGDTRr1gxDhgyp8TNt27ZFw4YNLQ0TeXl5iIqKQklJicUaNXHhwgXY2tpCURQIIardHhcm\nTpwIIgIRoVevXhgwYAAGDBig7fv1118t1o6OjsbChQslRvtweOedd+Dk5KSdP0VRoCgKIiMjpeir\nZTt37lwpWkIIeHt7Izo6Gl999ZX+AMvxww8/YMqUKSAidO3aFe+8845UfRnExsZWqV8tW7Z8aMd3\ncHDAm2++adZ3zDJMV1dXLWkaNmwIX1/fajf1eY7mkpaWpj1XEQA6d+5sWEP15ZdfQmZ/wcrKStuE\nELCyskLbtm2RlJRksea+ffsQGxtb7XuLFy9GTk6OxdrluX37Nrp16wYiwu7duy3WedC5OnXqFMLD\nw3Hr1i2L9NPT07WGJiIiokKuWMq1a9fw+uuvAwASEhLg7OwMe3t7DB06FPv27cO1a9d0H+O3337D\nhAkTtLqjbnXr1sXEiRNx584ds/QSExNBRDV2ymJjY0FE2Llzp9mxXrlyxdCOwf79+6uY2ujRoxEY\nGIj169dbrHvq1CnNIBctWoSSkhLMnDlT2ycD9bzJICcnB66urrh06ZIUPZVr166ha9eucHV1xd69\newEAly9fhpeXF+7du4dNmzahrKzMLM3KeVt+UxQFfn5+2LNnD86dO2eWbnBwcIVcO3ToEJo1awYh\nBEaMGCGtU5ycnIy0tLQq+4nI7GfhmnX2Dxw4gO+++w6LFy+Gs7NzlcJr2bIliAgxMTFmBVETNjY2\nEELA19dXil55unfvDldXVyla6olXNwcHB7i6uuoy+99++w1NmjSp8cHDvXr1kmaYx44d090YZGdn\nQwiBf/3rX9W+f+rUKbRo0UJXo/jqq69qhimEgL29Pd577z2L9QBUO4Jx6dIl3Lt3T5euyq+//opG\njRqBiODs7IyIiAjExsaiR48eICK0atUKK1asMEvz4MGDDzxXRGTRlVVUVFSNOZuQkIAtW7YgOTnZ\nbF0V1cB27NiBffv2Yd++fdi4cSOaNGmCAwcOWKR56tQpDBgwAIqioE2bNtq5KykpgaOjo3TDlHGF\nCQDz58/HiRMnpGipzJgxA0KIKp2wCxcuYOfOnRBCmD2CtGbNGsycORMff/wxFi9ejF69eqFXr15V\n2n8bGxuzdNW6XJ4jR45o9Xv48OFm6dVEXFwctm7dWu3xf/rpJ7O0pA7Jbt26FUIIKT3/uXPnQgiB\nV199VUJkFVm3bh2ICD179tSt1aJFC+3Ksm3btigsLKzy3pIlS8zuLdnZ2aFjx47VvhcdHQ0hhLQe\nmJ+fH4gI9erVs1hj5MiRGDBgQLUxFRUVoXHjxrquXHbs2KGZRHFxMdzc3LSKakm+5efnIzY2Fs2b\nN8fNmzctjqsmiouLIYTQGtiCgoIK7586dQp79+61qKPStm1b2Nvb1/oZIkL79u3Njrtly5ZwcXHR\nXv/0009o2bKl1hF88cUXQURwcnIyW3vo0KEYMGAAUlJSKuzXexUYEREBRVHQr1+/at/v3LkzvLy8\nLNZXkWmY6hVm+eH/l156SffQrBACvXr1qvY9e3t7Q0YPiouLMW7cOBARZs+ebfL3pkyZosXzySef\nYNOmTdi0aRMmT54sbWTx1q1bcHZ2rrYD3KRJE7P1pBpm8+bNpQxZhIWFwcbGBmPGjJH+0y4AEBMT\nAyLCN998o1tLNcs+ffpUGb5btmyZ9v6FCxfM0lUUBR999FG17zk6OqJu3boWx1wZtSF4UCNcGyNH\njoQQAr1798bevXu1LTY2Fn5+fhBC1Dq3+SA2b95coQKpV/VEZFGOrF+/HkIIQ34qDQC+++67ao2l\ntLQUqampaNWqFezs7EBEGDlypFnaRIRBgwY98DOWGKaiKIiKigJwf3hWNTNnZ2etrNSrOXMYN24c\nhBA4ffp0hf0lJSUQQiAwMNDsWFXatm0LRVFqHL1YtGjRY2eYnp6eEEJg5MiRWLVqFby9vbUpr6Sk\npAodb3MQQmDXrl3VvqeO2BnB+fPnQUS4ePGiyd+Ji4uDEAJnz56FtbV1tWsGvvzyS10d2qNHj1br\nSbm5uRg2bJjZetIM8/DhwyAi/Pjjj7q1jF5gQUSIiIiQolVbrKmpqejevbvZwwvZ2dm1fl5RFItO\ndk2oZqnXPL7//nvMmDEDjo6OmDFjBmbMmIFTp05BCIFRo0bp0lYN8+zZs9iyZQusrKzQuHFjEFGV\nRtgUHBwcIITAG2+8gZMnT+qKrTqKiopga2urmXpN8z9Xr141W9uUKQo9hrlkyRIkJyejcePG1c6x\nRkdHo3v37mbpenl5VTDZ1NRUTJ8+XTPj3Nxcs2MtH3NtBr5o0SIoioJ9+/ZZfAxA7hxmbagGau76\nh8LCQggh8Msvv1TYX1JSgu+++w5CCPy///f/ZIaq0aBBAxARbty4YdH3d+7cidDQ0BoX23Xu3Nmi\n30hu06YNmjRpgrFjxyIkJETLFSEEFi9ebLaetLMfGxuLvn376h4m/Prrr3X3OB+EJZO9NaFeQVaH\npVeYRUVF6Ny5c7VDjepcoaxVuGpHp3Xr1lL0KnPx4kUQke751j///BP29vbavEe/fv1w/vx5eHh4\nYMKECWbrlZ8LtbKywvr16/H222/j9OnTOH36NLZs2aIrXuB+L/a1115DWFgYpkyZos1bqpt6JWcO\nBQUFhhqm2pCow2LV8corr8DPz88sXXXY1MvLC15eXnBycqqw6EcPD+pgL1q0CEKIJ8YwgfuLdxwd\nHc0aor1z5w6EEHj33Xe1fVevXsXChQu1MqpsprKQUTalpaXIzc3FuXPnkJycjNzcXEydOhUNGjTQ\n6rylcSmKgo4dOyI0NBShoaGwsbGBjY0N4uLizNMzO4JqKCoqQteuXXHkyBFdOtevX0ePHj0ghEB0\ndLSM0KqQmZkJT09PaXo1GWZOTo42h9miRQuzV8MpigJfX19s2bIFW7ZswZw5czB8+HD07Nmz1uFa\nc1HnBseNGydFrzKjR4+WNlqwf/9+rQIUFxcDuN9Rc3V1NXvIe9q0aTX2ZtVt6NChUuJWGTlypBb/\nmjVrUFpaarbGp59+arJhdu7c2Wx99QrzpZdeqvGqTQjxwCHhyhQVFSE0NFTr4SuKgp07d2LIkCFI\nTEw0O87KMf83XWGqeHt7w8HBwazvqMP8K1aswJQpU9C2bVvY2trC3t4erVu3lrK+pDIpKSkgItja\n2lqsUdtq9PILgczF3d0d9vb2yM7OrrDfxcUFdnZ2Znf8dJ/9nTt3wsrKCo0aNdIrpY1jWzJMZSoT\nJ06UasbVGWZwcDDatGlT69WnKZw8eRJffvklvvzyS23fqFGjpK36S09PhxCixsVFeomMjJQ6/F0T\nQogKC1XM5cCBA1pHrfL24osvSonxk08+0RpcPeX9IMNMSkpCREQEevbsibt375qlrc5ZLlmyBMD9\nqxNFURAQEIBbt25h586d6NGjB9566y2L4y+PeoWplwcZZqdOnfD+++/rPo7sVbKmMH78eLO/U1JS\ngsTERG0FbnJycq1zm3pQ14O0bdvW4pXlGzduhI2NDRRFQUJCQrWfOXbsGKysrNC7d2894QK4v+rX\n0o6PbsNs06aNtEZRNcxTp07h2rVr2lZSUoKSkhJcu3YNKSkpiIqKQlRUFCZPnmz2MZycnBAfH687\nVhW1Yd29e7c2zFR+uE/2XOwbb7wBRVHw22+/6dbavHkziEi7D1E2LVq0gK2trfSl85WJiIjQXc7v\nvPMOhBCYMGECkpKSMHz4cGmG+dNPP6Fhw4YgItjZ2eHw4cMWa8XHx9dqmMOGDQMR4ejRoxbpCyEQ\nGhqqvd65cycuXLiAy5cvQwih6yqiPKmpqRBCSOloP+jWkRYtWmgjEnp42IZ55swZNG/eXLdOfHy8\ntLsXyvPFF19o7Z3e9kgdNq+pbqxZswaKomDq1Km6jqMe65EYZnJysrTVpgAQHh7+wGEydez53Xff\nrTBWbwrR0dEgIouGwmqi/G0l5R9c0KJFC1y4cMHi1W41oRqmDMLCwkBEhjw1B7jf+Kr3Asq+Qbsy\n7733Hl566SWLn1JUVFSEYcOGVbm6rHw7iCXUq1cPRGTRXGt1EBG6deuGkJAQbWvWrBmICA4ODhg7\ndqzF2levXoWHh0eFoVN1u379upT4AaBRo0YQQui6L1fl4sWLGD58uHZlXJ4mTZrU+LQsc+jdu7dU\nwxRCYOLEiVX25+TkYNWqVQgKCkKHDh10r7W4dOmSdjuJuSMOtZGbm6uVhyUXLtXRpk0b2NjYwMrK\nCvXq1dO28nVSRg4+EsNMS0tD69at8d5775n95IjaWLhwIebNm4d58+ZVKCh1nx48PT2lz0H8+OOP\niI6OrmKYshblVGbmzJnSevkdOnQwdE5GCIFXXnkF69ev171K9kHk5OSASN8j4bKystCiRQsIIdCm\nTRspceXn51eZd9VL165da1x1K+MRaLm5uZg2bRoaNWqEadOmYdq0aTh79qyEyP+DOhxbfrpBD6mp\nqWjUqJH2ZKzk5GSsX78eiqJgxowZurR/+OEHrYx/+OEHCdH+Z9HZhAkTMH78eO1iQT1O+/btpTxl\nSr26lDnSde/ePXzwwQcgInh7e0vtcK9bt04bqSu/9ezZU3f7rxIeHv7wDVOdXLb0MXiMZZQfLtPL\nlStXpD1nszoCAwMxd+5cZGVlGXaM8rzwwgvSOhOy6Ny5M4hId6P934b6CDuZ3Lx5Ez/99JNmkjNm\nzJB6VfUkMnjwYAgh92lp6u1czz333BNZvu+99x7CwsIs+q4AALIAIQQREZ09e5Y8PT0tkWAYqdy6\ndYu6dOlCqampjzoUDRcXFyopKaFffvmFWrZs+ajDeWyoU6cOLVy4kKZNm/aoQ/mvRvz/v7AyceJE\n+vjjj6VpEhFlZGSQk5OTFM0nBYsNk2EYhmH+l3h8fuCPYRiGYR5j2DAZhmEYxgTYMBmGYRjGBNgw\nGYZhGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZ\nhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgTYMBmG\nYRjGBNgwGYZhGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZh\nGMYE2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgR0GeaOHTsoJCSEevfuTdOnT6eSkhIp\nQWVnZ1NkZCT16dOHBgwYQMePH5eia5R2RkYGdejQgYKCgrRtxowZEqKtyMGDB8nDw4MyMjKk6CUm\nJlJ4eDj179+fIiMjKSsrS4runj17KpRFUFAQeXh4UEFBgRR9IvllQWRceRAR3b17lxYsWEAeHh5S\ndffu3UthYWEUFBREERERlJKSIk37SYyZSH5uGJnP8fHxFBYWRsHBwVLLwug6aFRuGFUeRJJihoUk\nJyfDx8cHV69eRVlZGWJiYvDhhx9aKleBMWPGYO3atQCAxMRETJ06VYquUdrp6ekIDAzUrVMbRUVF\nCA0NhY+PD9LT03XrFRYWwtfXF7///jsAIC4uDuPHj9etWx27du3C5MmTpenJLgvA+PIYN24cli5d\nCnd3d2RmZkrRvHLlCnr06IGMjAwAwLp16/Dyyy9L0QaezJiNyI3KyMrnrKwseHt74/z58wCADRs2\nYOjQobp1q0N2HTQiN4wuDxkxW2yYcXFxFczm119/xcCBAy2V07h69Sq6d++OkpIS3VoPS/thGObC\nhQuxevVqBAYGSmkI4uPjMXjwYO11QUEBOnTogPz8fN3a5bl9+zb69euHS5cuSdOUXRaA8eVx8uRJ\nAJDawGRnZ+Pf//639jo5ORndunWTog08mTEbkRvlkZnP169fx48//qi9Pnv2LJ555hndupUxog4a\nkRtGl4eMmC0ekhVCUFlZmfa6fv36dPnyZUvlNM6dO0fOzs60ePFi6t+/P40YMYLOnDmjW9do7YKC\nApo0aRIFBQXR3/72N7p48aIUXSKi5ORkSkhIoDFjxkjTTEtLIxcXF+21ra0t2dvbSzmH5dm6dSt1\n69aNWrVqJUXPiLIgMr48unbtKkWnPM2bNyd/f38iIiotLaXt27fT888/L03/SYvZqNwoj8x8btq0\nKQUEBGivDx06RF26dNGtWxnZdZDImNwwujxkxGyxYfr5+dGRI0coJSWFSktLaePGjXTnzh3dAd26\ndYtSUlLI29ub9u7dSwMHDqTJkydTaWnpY6tta2tLoaGh9M9//pN2795N/v7+NGnSJCkxA6C5c+fS\nrFmzqG7durr1VIqLi8na2rrCPmtrayoqKpJ2jLKyMlq7di2NHTtWip5RZUH0cMrDKOLi4sjf35+S\nkpJo2rRpjzock5Ads5G5oSI7n8uTmJhIcXFxFBsbK1XXyJiNxKjy0IvFhunm5kazZ8+mmJgYGjJk\nCLm5uZGdnZ3ugOzs7Khp06bUt29fIiIaPHgw5eXlUVpa2mOr3bhxY5ozZw45OzuToigUGRlJ169f\nlxLz5s2byc3Njby9vXVrlad+/fpVOji3b98mW1tbacf4+eefqX79+tSuXTspekaVBdHDKQ+jGD16\nNB09epRGjx5Nw4YNo9u3bz/qkB6I7JiNzA0V2fmscuDAAZo5cyatXLmS3NzcpGobFbORGFkeetG1\nSjY8PJy+/fZb2rZtG7m7u5O7u7vugJ5++mkqLCzUhnuFEKQoCimK/jtgjNLOy8uj9PT0CvvKysrI\nyspKly7R/VVj8fHx5O/vT/7+/pSZmUmDBg2io0eP6tJt27ZtheHG/Px8ysvLo9atW+sNWePgwYPU\nq1cvaXpGlQXRwykP2Vy8eJESEhKI6H4uh4aGUmFhIaWmpj7iyGrGqJiNzA0V2flMRJSQkEDz58+n\ntWvXUqdOnaRqExkTs5EYXR66sXQCNS0tDQMHDkReXh5KSkowduxYbNu2zVI5jbKyMgQHB2PTpk0A\ngN27d6NPnz64e/fuY6t9+PBhBAYG4s8//wQAbNq0CcHBwSgtLdUdc2VkLWYoLi7Gs88+i+PHjwMA\nli1bJnUVHXB/VZqMnKgJmQs7HkZ5AHIXSRw7dgz+/v7IysoCACQlJaFr167SF249iTEbsehHdj4X\nFRUhICAAp06dkqZZGaProMzceBjlAeiL2eJLoNatW9Pzzz9PYWFhJISgkJAQCg8P123gQghatmwZ\nzZw5k1avXk1NmzalpUuXSrlaM0q7Z8+e9Ne//pUiIiJICEGOjo60fPlyqlOnju6YjcLGxoaWLFlC\nb731FhUXF1OrVq1owYIFUo+RlZVFzZo1k6ppFEaWx/Xr12nEiBHa65EjR1KdOnUoLi6OHB0dLdbt\n3r07RUVFUWRkJJWVlVG9evXo/fffpwYNGvxPxmw0svM5Pj6ebty4UWUOd8OGDdKOY0QdNCo3jCwP\nWTELANAVCcMwDMP8D8CPxmMYhmEYE2DDZBiGYRgTYMNkGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNk\nGIZhGBN4bA3z559/JkVRqEGDBnTy5EndekuWLKHExEQJkTEMwzD/izyWhvn9999T//79acaMGXTy\n5EndP0ydnp5O//jHP8jPz09ShA9m//791KRJExJCaFtSUpLZOn5+fvTvf//bgAj/g/p4wPLb40zb\ntm2l6qWlpVU4T0IIGjt2LG3atIlk3Kbs4eHxWD/Eoia6d+8u7ddaPv300wrl++6779LZs2elaKu8\n+OKL9O2339b6GRk/iCCTZ555hqytrautgw4ODo86vGrJz88nd3d3OnTokFTdCxcuUGBgoJYjMh6E\nU5k2bdqQEILefPNNywRkPGqzWscrAAAgAElEQVQoLS0NMTEx0h5p1KRJEyiKIuVxeACwePFiSPpX\nH8jw4cPRuHFjCCHQrl07TJ8+HefOncMrr7yCmTNnmq1na2uL06dPS4+zoKAAiYmJAAAhRJWtcePG\nSEhIkH5cvSQlJVU5l9nZ2WjRooXFmqmpqdWWgRACy5cv1xsyevToAUVRdOuoLFmyBAEBARBCgIhA\nRHjppZfw1VdfSTsGAHh7e2PWrFlStKytrbVY1S06OlqKtgoRYcmSJbV+Ztq0aSbrdezYEZ9//nmV\n/UlJSXj77bfh4uICR0dHi9u9L774AlZWVhBCIDo6GidPnkR+fj7y8/Px8ccfQwiBDz74wCJtI7lz\n5w4CAwPRr18/aZqlpaWws7MDEaF58+YgIgQHB0vTVymffxZ939ID37t3D6dOnapSCRo2bAg/Pz/c\nu3fPUmkIIRAeHm7x9ytDRHBxcZGmVx3qST527Bju3LlT5f0LFy6YrfnFF1/gypUrFfbNnj3b4hhV\noqOjYW1tDSEEJk+ejN27d6Nz585QFKWCWXTo0MFkzZycHO3v9evXIzw8HESkNerlfxhWDw4ODlWS\nff369RBC6NJNTEyEEALDhg1Ds2bNKpSDDHbs2IGoqCgpWmqjsmrVKnh7e8PBwUErZ5msWrUKEydO\n1K1z/fp1hISE4OrVq9o+tb1YtmyZbv3ymrUZZmxsrFk/Vv3DDz8gKCgIiqJomxCiwmtFUfDZZ5+Z\nHau9vT2EELU+71YIgXfffdckvStXrsDLy0vL2aeeegoODg5o3ry5tgkh0LBhQ7NjrY4hQ4ZACIGD\nBw/q1uratStsbGy0tm7p0qUgImn1pTyPzDBnz56tHdje3h6KomivGzdubPHVYXZ2NhRFwb59+ywN\nrQo19WYTEhKwefNmKT1d1SBkMmfOnCr71q9fr1u3vBn4+vpq+0ePHo3o6Gg899xzEELA1tbWZM3o\n6Gjk5ORg1apVUBQFzZs3x6xZszBr1iytB62XW7dugYgwevToKsfWW/YDBgzQzDEjIwOxsbFaGd28\neVOXNnD/Ye6yGquJEyfixIkT2uuAgACtMZfJmTNnpGuqEBHq1q2LdevWSdFTO++1GWb37t3x9ddf\nm6VbUFCAc+fO4dixYzh37py2ubm5QVEU+Pn5mW0an376Kdq1a4eff/651s+ZY5gREREV6nV12i4u\nLtLOp0zDXLFiBa5du6a9Vg3zjz/+0K1dGdWjGjRoYNn3zf3CvXv3EB4eDkVR4OXlhXfeeQeenp66\nnVslMjISXl5eujQqQ0TYvHlzhX2DBw8GEWHx4sVISEioYByWHkNmj2jRokVwdXWtsn/QoEE4cuSI\nRZrbt2+vMuTYuXPnaj+rvm9K0jZo0ABCCMyaNQuXLl2q8n5hYaH2vqXcvHkTjo6OWLVqVZX31J61\npahDsvPnz6+w38bGBkIIfPTRRxZrq7z55pvSDFNl3rx5aN++vXbVY8SvUggh4O3tLU0vISEBDRs2\nBBFVO9xpCfn5+fDw8Kh1VGrz5s1SO7Tq1aUlXLt2DSUlJbV+5sCBA2jUqJFJeq+//jqEEBg4cGCt\nn7O1tZVmmD/88AOEEBg6dKgUvfJ4enoacnU5c+ZMzaPGjRtnkYbZGfT++++DiODp6YlNmzZVGZJ1\nd3e3KBCVFi1aVNso6qGyYSYkJFQxd72ViYjw/vvv69IoT8OGDTFv3rwK+86ePYu6detarLl58+YK\nZung4IAzZ85U+1n1M6b0yL29vR94zohI1/Bey5YtIYTA9evXq7zXqFEjXQ3Bu+++C19f3wpDhgDQ\nt29fCCHw97//3WJtldGjR0s3zPJzmBMnTqwwLC4LRVGkGeadO3e0uUwnJ6cac89cxo0bByLCL7/8\nUuNn+vfvL80w161bp8swH0RGRgYcHBzw8ccfm/T5uXPnYuPGjQ/8HBFhwoQJesMDcP8nDYUQ6NSp\nEwoKCqRoqvTo0QO5ublSNQsLC9GrVy+tvhw9etQiHbMzyM/PD0SEp556Cvb29rhx4wZ+/fVX2NjY\nwNfXF3l5eRYFotKiRYsK8wMdO3bU/jZ1eKIy6pVk+deVryiruwo1lfz8fHh7e+PUqVPaVlxcbJEW\nAOTm5iIkJKTa9yIiIizWVecU1U39/c7qMMcwTT22pYaZmZmpxT5ixAj88ssvFeZ29cw1RkVF1fjd\nL774QpphHjt2DESEQYMG6dYqz7x58yrMFRMR9uzZI02/Q4cOEEJUGAK2hAsXLlTpXKtxP2hosjby\n8vJMGtmSMfpVXFwMPz8/KIoCa2tr3WVSHQUFBdKv6oH77YaetqM61JENmR21KVOm4LvvvpOmp1J+\nvU2TJk0s7qyZnUG5ublYv349zp49C+C+Wajzd8nJyRYFUZ4WLVqgUaNGCAoKwr59+3D79m3s27dP\nV4+uuqvJ8gYKQNeioL1791ZpCLp27WqxaRYVFcHBwQEzZsyoMLYP3F8hmZaWhrS0NLMXVpU3zE6d\nOlX7mZYtW2pDkUIIaT+CLYSw2DBzcnIqxE5EcHR0RHh4OLp06WLScFRNODk5PRTDBO6XgZOTkxSt\n8pw4cQKzZs3Shmd79eolTdvb21vK6El6ejq8vLzg5eWFr776Cvv370dgYCCICM7OzkhKSrJI9/z5\n8yCiBy4SlGGY/v7+mknIHgUD7pvl2LFjYWNjg2+++UaabkpKCqytrfHrr79K0wT+057IMszc3Fz4\n+PhI0apMecN84YUXLNbRPUaxbt06KcmoMnPmTIwfP77Kfk9PT4sN09fXt4phlr+a3Lx5cxUDNYeU\nlBQMGTIEsbGxGDlyJDp06AAiQkREhMWm+fnnn2POnDlwc3NDx44d0aFDB3Ts2BHOzs6aId++fdss\nzfKmU3nucs2aNdrVlrrZ2dlZFHtNx9azYGnUqFHo2LEj3Nzcqpinra2txYvEars6fVIMU6V8x0IW\nI0eO1NXZqY0jR45g6NChICK0atXK7O/v2bMH3bp1Q7t27XD16lWUlZVpt2Xcvn0b+fn5KCsrAwCT\nTLU2bt26peWbEQuhVLMUQmD//v1Stb/55hsIIVBUVCRVV/YV5vDhw2u8cCkoKMD8+fNrHXavjfKG\naelIIiDBMNUg1qxZo1cKAPDRRx+hY8eOVcxAXRZtKeq9mOULTV0IIHt1K3B/HoKIsGLFCqm6ERER\nFo/vjxgxosb7DStvAQEBuHz5srS4ZQ8xqYwfPx7t27e3+PumGKas+RSjDfP111/X3TGpzKpVq6Rf\ntVZGHaFKSUkx63vqrVzVbe3atUOrVq1ARNpV8rFjxyyKb+PGjRVuKzE3zuq4dOkS/vjjD8TExGj1\n0sXFRcooXXnUW66EEPD09ISnp6c0bZmGefv2bTg5OVUp29deew329vZwcnLStSI3KysLnp6esLGx\n0eVVupzi4sWLICKEhoZKG7oD7i80eOWVV7TXly5dkjLJrt5+UH7z9fWVagwAcPDgQXTu3BnPPfcc\nbt26JVVbzzxEdatkq9scHByQmZkpLeacnJwqK1BlMXjwYMyYMcPi76v/c2pqapX31NWHshBCoFmz\nZrXee2cJOTk5iI6OhhACzZs3r3alsqWsWrXKsKsqlRUrVoCIsHPnTrO+t3z5chARWrduDR8fHwQG\nBmLevHnYs2cPbty4gYyMDOzbtw8DBgyAl5eXxbcH+fr6QlEUdO7cGUuXLtV1jzlwfxpLXbFaeXvh\nhReQlZUF4P4tdmPHjrX4ONevX6/2GLKQaZgzZ85EfHy89vrYsWPakL2fn59u/YyMDLi6uj66Idlv\nvvkGRIQ33njD4oPXxIkTJ+Ds7FzhRmGZE+yqWQ4ePFjK02xu3bqF+fPnY/LkySAi2NjYGFIuwP0e\n48WLFy3+fnJyco1GaW1tje+//15itPdRb7KXzeHDh0FE2L59u8UaY8aM0f7/evXqVZi/FUJIvbFe\nzeV//etf0jTVqxNFUfDSSy9J01U5c+aMIYaZn5+PgwcPVujEGvVkqZycHBQWFpr9vRs3bmD58uVS\nV8SqDwwpP+2hroZdvnw57OzstPcq329sLgsWLEBAQAAaN26M8+fPS4i+IrIMc+7cuSAizJ07F46O\njrCyskK/fv1w+PBhSZH+5wrzkRmmvb29rmGOB/Hzzz9DURQEBgZavDq2JqKjo+Hi4oLBgwfrvrpc\nuHAh6tatqy2X9/X1xcmTJyVFWpVr167h0KFDujQ6depUxSwDAwN1zePWhhFD3gDw2WefgYh0LWZY\nt25djR2IhQsXSh05kWWYBQUFCA8P1+bUWrdubcg9mCpq3DI4c+YMTp8+jeeee67KEOrjxqBBgzSz\nbN68uRTNunXravllY2NTxWxycnIQFBSEoKAgKcdTbwE0AlmG2aZNmwq5sHfvXkkR/ofMzEy4u7s/\nGsNcsmQJiEjX8zv/Wxg2bBj8/PzMXoCj95hPCuptD0ahJ/lVcnNz8cUXX8DT0xNr1qxBSUnJA28s\nf5SMHz9eu/VA9vNjqyMpKUm3Yebm5mLRokVao+jk5IRFixY98Nmvj5K6detCURS0adMG586de9Th\nWIQQpj8tyFw6d+6M4OBgac/8fhIQgPk/x7Bs2TK6efMmvfLKK9SyZUtzv878j3Dt2jXy8fGhoqIi\nys7OftTh/NewYcMGWrNmDR08ePBRh/Jfzf/93//RvXv36N1336WwsLBHHY5FKIpCCxcupOnTpz/q\nUP4rsMgwGcYUTpw4QT4+PpSdnU3NmjV71OEwDMPogg2TYRiGYUzg8f6lYIZhGIZ5TGDDZBiGYRgT\nYMNkGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNkGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNkGIZhGBNg\nw2QYhmEYE2DDZBiGYRgTYMNkGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNkGIZhGBNgw2QYhmEYE2DD\nZBiGYRgTYMNkGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNkGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNk\nGIZhGBNgw2QYhmEYE2DDZBiGYRgTYMNkGIZhGBNgw2QYhmEYE9BlmHfv3qUFCxaQh4cHZWVlSQlo\nz549FBQUVGHz8PCggoICKfoqBw8eJA8PD8rIyJCiFx8fT2FhYRQcHEwRERGUkpIiRZeIaO/evRQW\nFkZBQUHStDMyMqhDhw4VynnGjBkSor2PEblBZGw5q8jODSN1s7OzKTIykvr06UMDBgyg48ePS9E1\nIueIjMs7o/PZqHJOTEyk8PBw6t+/P0VGRkqtKypG5J1R9XvHjh0UEhJCvXv3punTp1NJSYluTam5\nAR2MGzcOS5cuhbu7OzIzM/VI1ciuXbswefJkqZpFRUUIDQ2Fj48P0tPTdetlZWXB29sb58+fBwBs\n2LABQ4cO1a0LAFeuXEGPHj2QkZEBAFi3bh1efvll3brp6ekIDAzUrVMTRuSGkeWsIjs3jNYdM2YM\n1q5dCwBITEzE1KlTdWsalXOAcXlndD4bUc6FhYXw9fXF77//DgCIi4vD+PHjdeuWx6i8M6J+Jycn\nw8fHB1evXkVZWRliYmLw4Ycf6taVmRu6rjAnTZpEU6dO1SNRK3fu3KGlS5fS9OnTpeouX76cBg4c\nSLa2tlL0rKysaPHixeTm5kZERM888wxduHBBqraTkxMREfn5+VFqaqoUbSMxIjeMLGcV2blhpG5m\nZiadPn2aRowYQUREvr6+tHTpUt26T2rOGYVR5Xz06FFycXGhDh06EBHRyy+/TEeOHJE6mmZUPhtR\nv48ePUq+vr7UsmVLEkLQ6NGjad++fVKPoRddhtm1a1dZcVTL1q1bqVu3btSqVStpmsnJyZSQkEBj\nxoyRptm0aVMKCAjQXh86dIi6dOkiRbt58+bk7+9PRESlpaW0fft2ev7556VoFxQU0KRJkygoKIj+\n9re/0cWLF6XoEhmTG0aWM5ExuWGk7rlz58jZ2ZkWL15M/fv3pxEjRtCZM2d06xqZc0TG5Z1RukaV\nc1paGrm4uGivbW1tyd7eni5fvqxbm8i4vCMypn4LIaisrEx7Xb9+fWllISs3HttFP2VlZbR27Voa\nO3asNE0ANHfuXJo1axbVrVtXmm55EhMTKS4ujmJjY6XqxsXFkb+/PyUlJdG0adN069na2lJoaCj9\n85//pN27d5O/vz9NmjSJSktLJURrPLLL2ajcMDLnbt26RSkpKeTt7U179+6lgQMH0uTJk6WdQ9k5\nR2Rc3hmZz0aVc3FxMVlbW1fYZ21tTUVFRbp0iR5OWycbPz8/OnLkCKWkpFBpaSlt3LiR7ty5o1tX\nam7IGNc1Yg4zKSkJoaGhUjW/+OILvP7669rrwMBAqeP6+/fvR0BAAH777TdpmuUpKyvDzp07ERgY\niOLiYuna3bp10+YHZWFEbhhRzkblhpE5Fx8fj4CAAO21EefQyJxT9Y3IO5m6RpXz2rVrERMTU2Hf\ns88+izNnzujSBYxv61Rk1+9t27YhJCQE4eHh2LBhA3x8fKRpq+g5f4/tFebBgwepV69eUjXj4+Mp\nPj6e/P39yd/fnzIzM2nQoEF09OhR3doJCQk0f/58Wrt2LXXq1ElCtPe5ePEiJSQkENH9IYvQ0FAq\nLCzUPaeUl5dH6enpFfaVlZWRlZWVLl2jMaqcjcoNI3Pu6aefpsLCQm0YSwhBiqKQouir1kblHJFx\neWdkPhtVzm3btq0w5Jifn095eXnUunVrXbpExuadkYSHh9O3335L27ZtI3d3d3J3d9etKTU3ZDi2\nEVcR48aNw7Zt26RqVkZWr6uoqAgBAQE4deqUhKgqcuzYMfj7+yMrKwvA/Svvrl27Ij8/X5fu4cOH\nERgYiD///BMAsGnTJgQHB6O0tFR3zOWRmRtGlnNljOqRy9QtKytDcHAwNm3aBADYvXs3+vTpg7t3\n7+rSNSrnAOPyzsh8Nqqci4uL8eyzz+L48eMAgGXLlkm/I0DlSbjCTEtLw8CBA5GXl4eSkhKMHTtW\nigfIzA2Lu1/Xr1/XVo0REY0cOZLq1KlDcXFx5OjoaKmsRlZWFjVr1ky3zsMgPj6ebty4UWWeZ8OG\nDbr/h+7du1NUVBRFRkZSWVkZ1atXj95//31q0KCBLt2ePXvSX//6V4qIiCAhBDk6OtLy5cupTp06\nunSJjMsNI8v5SUQIQcuWLaOZM2fS6tWrqWnTprR06VLdV1VG5RyRcXlnZD4bVc42Nja0ZMkSeuut\nt6i4uJhatWpFCxYs0B2v0RhVv1u3bk3PP/88hYWFkRCCQkJCKDw8XHe8MnNDAIDuiBiGYRjmv5zH\ndg6TYRiGYR4n2DAZhmEYxgTYMBmGYRjGBNgwGYZhGMYE2DAZhmEYxgTYMJknhm+//ZZatWpFQghp\nP6/EMAxjKmyYzBPDlClTaO3atbRr1y7q06cPvf322486pMeKixcv0lNPPaU9pYfRz82bN8nDw+NR\nh8E8Jug2zKioKBlx/Newbds2at++ve7HZmVnZ1OnTp20R3AdOHBAUoQVGTlyJE2cOJE2bNhAJ06c\n0DZLyczMpDp16ki5Ybwyqamp1LdvX/rLX/5C169fJxsbG2rTpo3048ji//7v/2p8T08ZV8ebb75J\n7dq1o1GjRtGzzz4rVfths3PnTjp58qTF33d0dKSwsDDdPw3VrVs3atKkCQ0cOFCXTm0UFRVReno6\nrVy5kp5++mlSFEXXw9c3bNigu+0xh507d5IQ4qEd75Gj97FDLVq00CthOG+//TZq+leJSPpD3tPS\n0uDo6Ihr165ZrDFu3Dg0bNgQM2fOhLW1Nfr16ycxwvvk5OTA1dUViqJACAFFUbS/v/rqK4t1VZ1b\nt25JjLYqmZmZcHNz061z9+5dxMbGVvmfP/nkEwwdOhTXr1+3SFdRlFrf01PG5Tl27BhsbGwghEBR\nUZFFGocPH8bhw4fRuXPnClvHjh0RHh6Ow4cPS4m1Ou7du4eDBw9i0KBBsLOzQ/fu3fHpp5+a/P3j\nx49jypQpiIqKwsCBA/Hee+9JiUsIASGEFK2a2LJli1Zf1G3+/PkW1x1/f38oilLr4+patWql/Ri2\nXrp161Zj2/okUVxcjJEjRz7wc7r/UwcHB4sraWXS09Mxd+5cPPXUUyAi2NjYwN7eHkSEuLg4i3U9\nPDxqNcxGjRpZrF0dZ86cQYMGDaTp3b59G0IIHD16VJomcP8ZoTU16ocOHbJY19raGoqiIDo62mIN\nUzl16hS+/PJLi7+fmJiIFi1aQAgBe3t7bX9KSorWYH7wwQcWaSuKUuMvTyiKAldXV4t0K0NEaNiw\nIS5dumSxhvq/Vt6ISPt7wIABumO9dOkS+vbtC0dHR3h6esLBwQENGzbEmDFjtGd9mou1tTV+/PFH\n3bGVp3PnzhBCICoqSqpuZWxsbKAoClatWoU7d+4AuJ8bQ4cOtUhPNd2a2uSYmBjteDIgImmGefjw\nYfTv379CzgkhEBkZKUW/PH/++ScGDRoEV1dXNGnSBLt27TLpe7r/UyLCTz/9pFcGGRkZmjm6urqi\nW7du2Lt3LzIyMuDp6WlWj7MyrVq1qvakbtmyBUQET09PPaFX4bvvvpPaM01MTDTMMI3oQQcEBEBR\nFLRs2RLnzp2Trl+ZiIgIi78bHBysVcxXX31V21/eMGNjYy3SFkLU2DCpZqQXtTP18ccf69IZMGBA\nhe3y5cu4fPky1q9fj0aNGkkzTHd3d62RtbKyQps2bXDx4kVdmoqiSDdMtRMi+2fHKqPmWPlOn5eX\nl8X1UjXM6sjNzYW3tzcURZHWlhARrK2tdeucOHEC1tbWEELA3d0dfn5+eOONN2BnZyf1CvaHH35A\nbGysltOjRo1CXl6eyef5sTFM4L7R/PHHH1X279y5E3379rVYt7pe0JQpU7T977zzjsXalQkPD0fr\n1q11DcdWJiQkxJAh2fDwcEMMs7CwEM2aNYOiKHBxcUFeXp70Y6gUFBRYbJgFBQUVerKZmZm4e/cu\ntmzZgu7du2v7LW00TTHM48ePY9asWRbp9+7dG7GxsboNpyYyMzMr9PYvX76sS69v374gIvj7+0uK\n8D5EpBnm3bt3cevWLaSmpuLo0aMoKyuzWLNZs2YV9u3fvx8hISFo0qRJhbzp27evxVf31RncV199\nVetwvrl6KpcvX4aiKDh9+rRF2pUZMWIEiMji/FVR60J1nZ7g4GAQke7pnRYtWqBFixYYM2YMtm/f\nbrGOFMP87LPP9MrUSlxcnK6hkcqG+cYbb8DKygpEBHt7e2RnZ8sIE8D9hDVlLNxUzp8/D0dHR/To\n0QOJiYkoKCiQpq3OW3p7e2ubLN5++21tPvSbb76RpluZ77//3uL5mPJXkUIIrF69WhuKK7/dvn3b\nIn1FUWoc2lYUBe3bt0dAQAAmTpxokb6tra30Dk9GRgY2btwIR0dHrXevlsOyZcssbmwSEhLQqFEj\njBgxAvfu3ZMas9rYbt26FZ6ennBxcdHqfHBwML7//nuLNCsbpq+vr1YWvXv3xt///neEhYXBxsYG\nrq6uePPNN806xpUrV6o1uISEBEMMc8eOHVAUBRkZGRZpV6ZBgwYgIixbtsxijXXr1mnnqTqcnJxA\nRHjjjTcsPsb58+fx9ttvS/m/pRimkQ3inTt34O3tbXFPEQAaNmyoGWZGRoZWmYhI2hxFTk4OvL29\nceLECSl6KikpKejXr1+FBrywsFC37pkzZyCEwOzZs7V9165dg6Ojo5RffAf+03OU2ajv2LGjwlVb\nRESExbnx0Ucf1Tpnp149WEpNc0VnzpzROhN6RiKEEPj666+Rm5uLmJgYeHp6wt3dHd99950uzdrK\nQ93M/d1GOzs7WFtb4+LFi1i+fDnatWuHgQMHSulsd+vWDW3btq12dAoAli5dimPHjpmlSUTo0qVL\nhX3q/15dJ8iSBULR0dFSDVNdQBQYGIiYmBht27FjB37++edazdQSiAhz587VpTF37lx4eXlh3759\nSE1NRW5uLl544QXY29tXyMGSkhKL9NXRIlk8VkOy1bFq1SrdY9ijRo0CEeGTTz6Bra1tBcPcu3ev\nlDhff/11rcKkpaUhKSkJSUlJWLJkCZYsWYIJEyboMrrPP/8cPj4+EELAy8tL94+2TpgwAUFBQVX2\n1zaMaC7lV9/q5dq1axgzZgxsbGxgY2ODBQsW4PDhw7rMuPIVZnUGERISYrG+qqfmwuuvvw5PT88K\nx7KUQ4cOwcbGBj/99BNmzpyJCRMmICIiAkIIXYuJ1Hoxbtw4zJs3r8J7UVFRaN26tUULPdS5LvWq\nxNXVFY0aNQIRScu3migqKjJ7NX91V5gPMkxz10K0bNkSrVu3rrLAx1LD3Lx58wMXbsnsvMqYW1Rj\nJiI89dRTcHV1rRKvnuPk5eWhcePGuHDhgu5YAQmG2bx5c4uHrB6Eq6srunbtCiKqsfdoCvv370dY\nWJhW0YcNGyZ1dRdQcXizefPmFW7RUN9bv369rmPcvHlTa3D1rIgE7i/4qe7qRuYKuvKGqffX6fv2\n7asZQUlJidbwenl5Wax54cIF2NnZISoqCnZ2drCzs6vSoXrxxRct1nd0dKxw/ivfvtO+fXuLtXv2\n7FnFzPPz83U3iqdOncKpU6dq/DX6kydPWnQMtTxfffVVHDlyBMD9jmV0dDSaN2+OFStWWByzKdSt\nWxdnz541+fPVLfpR/+/hw4dX+bwQAps2bTIrJiFEtathLTVM9XvlN09PT3h6eqJBgwZSrzALCwul\ntZ/Ozs4VTPK1117Dd999h48//hhCCDg5OVmsff78eYSEhKBTp04ICAjQHetjfR9mTEyMVtE6d+4s\n7QR5eXlJMcx58+ZpjWCvXr3wwQcf1HoVaU6FrcyVK1cwZ84c2NjYwM3NzeJ7A4H7iwqqu7oE5Brm\nBx98oBlEr169dGkREVJTU7XXtra22nxVbm6uvkArMXXqVK3y6p1uuHTpEk6cOFFlqF5vOfv7++O1\n116rsO/atWsgIotX9fSH3fwAACAASURBVKo8aEhXvedTFrI7r9VhbW2NGzdumPz5u3fvVtsxWLx4\nMVq2bAlbW1vNIE+fPg0hBLZt22ZWTEKIaoekhwwZInXoFAB2794NRVHw/vvvS9ELDQ3F0qVLpWjV\ndgwfHx9pF2RRUVG6b4HUnaUdOnTQK1EjL7zwArp06YI9e/Zg165d6N27txTd2bNng4h03fSurjAV\nQqB169ZS4qqN8qs2U1JSdGl5e3tXa5g5OTkQQkibw1Rvr1EUBfb29khLS7NYi4hw4sQJJCYm4rPP\nPoOnpycuXbqE06dPo3v37tpVi16KiorQrl07KUOytSGE0LXIqmfPnggMDMQXX3wB4P5V986dO1G3\nbl1dUyQHDhxAw4YNaxwRUM+pLMP8448/QEQICwuTolcTNjY2Zn9H/T+joqIqXGkeP34cwcHBeOqp\np/DRRx/ByckJ69atM3suXQiBxMTEKvv9/PykG2arVq2kGqabm5vUBYiVSU9PR7NmzbBlyxZpmkSk\ne8W+LsO8ffs2vv76a10B1ERubi6GDh0qfUUdcP8pOkRUZY7GHFatWmX43AsA7d4kIQTmzJmje2gT\nuL9Iovz/fubMGa0DIPN2GAB46623tKEgPz8/i3Wys7MxY8YMxMXFIT8/v8J7ZWVlWLVqFRo1aqR7\nPt3FxaXC/IlRT7gRQmDlypUWf7+4uLhCJ0od0srKytIV1/nz56vMs1aeExsxYoSuY+Tl5SE4OBh1\n69YFEWHHjh269Zo0aVJtWzFq1Cg0b97c4iuL7du3o1OnTtXODbZv3x6bNm2y+AooNDQUAwYMQG5u\nLq5fv474+Hj07t0bnTt3xtWrVy3SrI7Vq1ejXr16sLOzM/kG/dq4fPmy4U94U0d5ZNK2bVvdGroM\n86effkJycrLuIKojJiYGp06dMkS7X79+ICJpj9AyErVyzp49u8Z5JXMZOXIkXF1dkZOTg1WrVqFB\ngwYQQmDQoEFS9Mtz8eJFzTAXLVokXV82lQ3CKMOUMfRdUlKCI0eOwNPTEwEBAdi9e7eU2AoLCzFi\nxAiEhoZWKA8XFxeMGDHCrKFN4P7VQkZGBlauXImJEydqCzsaNWok7WlQixcvxsaNG7X7L7ds2YKh\nQ4fCwcEBc+bM0aV99+5dZGRkYM6cORU2vaaWnZ2tzTOqV4CKoiA9PV2XbmWWLFkCRVFgY2OjjUjo\nQX3gi1H88ssvFe5skEF4eLgUvcfyIYCjRo2Cs7OzYfrPPffcE2OYRnHp0iWT5l3/1yhvEFu3bjXs\nOC+99JKuxw8+aZw8eRLff/+99BGMJ53vv/8ejRo1wsCBA7Fy5UrcvHnTkONMmTJF1yMkKzNq1Cjt\nUX6yUevg9OnTpehlZ2ejR48eUp7aJACAHjOcnZ1p4sSJNGvWLEP0r1y5QsOHD6cBAwbQP/7xD0OO\nwTyZCCFICEGjRo2iNWvWkJWV1aMOiWH+pxBCkKurK61YsYKCgoJ065WWltK9e/fI2tpaf2yPo2Ey\nDMMwzOMG/4A0wzAMw5gAGybDMAzDmAAbJsMwDMOYABsmwzAMw5gAGybDMAzDmAAbJsMwDMOYABsm\nwzAMw5gAGybDMAzDmAAbJsMwDMOYABsmwzAMw5gAGybDMAzDmAAbJsMwDMOYABsmwzAMw5gAGybD\nMAzDmAAbJsMwDMOYABsmwzAMw5gAGybDMAzDmAAbJsMwDMOYABsmwzAMw5gAGybDMAzDmAAbJsMw\nDMOYABsmwzAMw5gAGybDMAzDmAAbJsMwDMOYABsmwzAMw5gAGybDMAzDmAAbJsMwDMOYABsmwzAM\nw5gAGybDMAzDmIDFhrlnzx4KCgqqsHl4eFBBQYHuoL766iv6y1/+QsHBwRQZGUmpqam6NVV27NhB\nISEh1Lt3b5o+fTqVlJRI0d27dy+FhYVRUFAQRUREUEpKihRdI8tZ5eDBg+Th4UEZGRnSNLOzsyky\nMpL69OlDAwYMoOPHj0vTJjImZqPOYWJiIoWHh1P//v0pMjKSsrKypOgamRtGxUxEFB8fT2FhYRQc\nHCy1nI2q20REd+/epQULFpCHh8cTcf6IjMtno84fkXHnUFpZQxK7du3C5MmTdetcuHABPj4+yMrK\nAgB8/vnnGDZsmG5dAEhOToaPjw+uXr2KsrIyxMTE4MMPP9Ste+XKFfTo0QMZGRkAgHXr1uHll1/W\nrVsdsspZpaioCKGhofDx8UF6ero03TFjxmDt2rUAgMTEREydOlWathExG3UOCwsL4evri99//x0A\nEBcXh/Hjx+vWrQ5ZuWFkzFlZWfD29sb58+cBABs2bMDQoUN16xpVt1XGjRuHpUuXwt3dHZmZmdJ0\nyyOzbhuVz0adP8D4c1geS8taimHevn0b/fr1w6VLl3Rr7d27F0OGDNFe//HHH3jmmWd06wL3K375\nhvvXX3/FwIEDdetmZ2fj3//+t/Y6OTkZ3bp1061bGZnlrLJw4UKsXr0agYGB0szn6tWr6N69O0pK\nSqToVcaImI06h/Hx8Rg8eLD2uqCgAB06dEB+fr5u7fLIzA0jY75+/Tp+/PFH7fXZs2el1G+j6rbK\nyZMnAcAww5Rdt43KZ6POH2D8OVTRU9ZS5jC3bt1K3bp1o1atWunW6tKlC12+fJlSUlIIAO3bt4+e\nffZZCVESCSGorKxMe12/fn26fPmybt3mzZuTv78/ERGVlpbS9u3b6fnnn9etWxmZ5UxElJycTAkJ\nCTRmzBgpeirnzp0jZ2dnWrx4MfXv359GjBhBZ86ckaJtVMxGncO0tDRycXHRXtva2pK9vb2UvCuP\nzNwwMuamTZtSQECA9vrQoUPUpUsX3bpG1W2Vrl27StOqDtl126h8Nur8ERl/DlX0lLVuwywrK6O1\na9fS2LFj9UoREZGjoyPFxMTQiy++SD4+PrRx40aaNm2aFG0/Pz86cuQIpaSkUGlpKW3cuJHu3Lkj\nRZuIKC4ujvz9/SkpKUlazCqyyxkAzZ07l2bNmkV169aVoqly69YtSklJIW9vb9q7dy8NHDiQJk+e\nTKWlpbp0jYxZRfY5LC4uJmtr6wr7rK2tqaioSLe2iuzceBgxE92fJ42Li6PY2FjdWkbXbSORff7K\nY2SbJPP8ET2cc6i3rHUb5s8//0z169endu3a6ZUiIqIzZ87QihUr6MCBA3T8+HH6xz/+QVFRUQRA\nt7abmxvNnj2bYmJiaMiQIeTm5kZ2dnYSor7P6NGj6ejRozR69GgaNmwY3b59W5q27HLevHkzubm5\nkbe3txS98tjZ2VHTpk2pb9++REQ0ePBgysvLo7S0NF26RsasIvsc1q9fv0qlv337Ntna2urSLY/s\n3HgYMR84cIBmzpxJK1euJDc3N916RtdtI5F9/spjVJsk+/wRPZxzqLesdRvmwYMHqVevXnplNBIT\nE6lr16709NNPExHRX/7yF7pw4QLl5uZK0Q8PD6dvv/2Wtm3bRu7u7uTu7q5b8+LFi5SQkEBE94cV\nQkNDqbCwUOrqXtnlHB8fT/Hx8eTv70/+/v6UmZlJgwYNoqNHj+rWfvrpp6mwsFAbXhFCkKIopCj6\n0s3ImP+/9s48qqqqf+PffWOUQVDGnLAIaAWCvo6vmWAqmhpimviaIS0pWRmlrxPLSpJl2ijKmzmW\nRpmImr2+as4DiQOZoQsVNGeZRFSQQRCe3x+sc7qXQe+9ex/Ffvuz1ll6zuU+53v32Xs/Zw9nH62u\n4VNPPWXQrVRaWkq3b9+mDh06cOnqIzpvaB1zeno6zZ07l7755hsKCAgQokmkTdl+GIi+fkTa1kla\nXT8i7a8hb1pzG+aZM2fo6aef5pVR6dixIx0/flw1yP3795Orqys5Oztza1+6dInCwsKopKSEqqur\nacmSJTRixAhu3eLiYpo+fToVFBQQEdGxY8eourraYByIF9HpvHz5cjp06BAdPHiQDh48SJ6enrR+\n/Xrq2bMnt7avry+5ublRamoqERFt27aNHB0ducdntIxZq2vYo0cPys3Npd9++42IiFatWkUhISHU\nokUL7pgVROcNLWOuqKiguLg4SkpKEhqzVmX7YSD6+hFpl5+1un5ED+ca8qa1BW8A+fn55OLiwiuj\n0q9fP8rKyqKIiAgiIrK3t6fExERijHFrd+jQgV588UUKCwsjxhgNGTKEwsPDuXW7detGMTExFBUV\nRbW1tWRlZUULFiwge3t7bm0F0emsJYwxWrRoEc2cOZOWLVtGrVu3poULF5KFBXd20wytrqGNjQ19\n+eWXNGfOHKqoqKD27dvT/PnzBUVdh+i8oWXMu3fvpuLi4gbjad9//z3Xb9CqbBMRFRUV0Wuvvabu\njxs3jp544glavXo1ubu7c+trUba1ys9aXT8iba+hAm9aM4gYHJRIJBKJ5G+OXBpPIpFIJBIjkIYp\nkUgkEokRSMOUSCQSicQIpGFKJBKJRGIE0jAlEolEIjECaZgS4dTW1tLy5csfdRgSiUQiFGmYgjhy\n5AgFBwdTWVnZow7lkZOTk0MTJ0581GFIJBKJUJq1YWZmZlJmZuajDsMoDh8+TIcOHSJHR0dyd3d/\nLFtYItayrK6upk6dOlFsbKyAiExj8ODBJi9wsW/fPoqPj6f4+HhijBFjjEJCQig+Pl5obCUlJUL1\ntMbS0lJND2Vpw8jISPr0009N1rpw4YK6NKJOpyNPT08KCQkhHx8fcnNzoz59+tDRo0c1+BXiKCgo\nIMYYff311486FKPYs2cPtW7dmlq0aEGMMXJ0dHzUIf09EPqiMUGUl5cjKSkJjDEwxhAYGIigoCAE\nBQVhz549KCkpMUuXiNClSxf07NkTU6ZMQXJyMmpqaoTF/fnnn0On06mb8vLWx4VnnnmGW+Onn37C\nypUrBURjGmvWrAFjDO7u7iZ9j4ia3Pbu3SsktsTERPj4+GDWrFlIT09HcnKyujVX7pcuS5YsMUnr\n5MmTalkOCQnBxYsX1c9ycnLAGMOwYcOExH3ixAkEBgbC2toaHh4eICJMmzaNWzcrKwuMMURFRQmI\nUntatWqFoKAgnD9/HpMmTUIzreqbZPv27ejWrRs++ugjoXU0Lw9MxdmzZ9+38MyePVtYxQIAhw8f\nVrXj4+MRHx+PtLQ0pKamYufOnfDy8jLr4o8ZM6bR4xs3bsSdO3d4wzagvLwcs2bNAhFhzpw5Jn3X\n2tparVyM3cwlJyfHIC15DXPLli2wt7dv9LP4+Hi0atUK4eHhZmmXl5cjNTW10c86d+4Mxhg2bdpk\nsu7s2bPVbe/evZg9ezaCg4MN8re55OTkoHPnzgblxdXVFT169EBcXBx69eoFInqgWXh5ecHCwgIW\nFhbw9PRU/x8UFITAwEC4urpi0KBBXLEqrFq1CjqdDkSEQYMGqcfv3r2LkSNHgoig0+kQExPDfS59\n7cGDB2PdunVcOgEBAQZp3apVK/X/vERGRgoxzNzcXOTm5qr7d+7cUY/l5ubi+PHjICLY2NhwnWfx\n4sWq0Vy4cAGtWrXi0gPqXtzep0+fBvWPn58ft7Y+Dg4OYIzhhRdeAGPM7AYSAHz22WdwdXVVt8bq\nzzZt2iAnJ8covQfmpL17997XMJUtODjY7B+lz7Rp08AYQ8+ePRv9fMKECWaZxLPPPssbmskwxtC7\nd2+TvpOWlobExESDbdq0aXBxcYGLiwscHR3h6OjYLA1z4sSJ6NWrV4PjaWlpaqvb2toaBw4cMFl7\n4MCBTVYiKSkp8PDwMFmzKfRvEnlMqEOHDiAi9OjRAwkJCejRowcuXLhg8DdDhgx5YIU+ePBgMMaM\nKoeDBg1CZGQk8vLyzIq5Xbt2qlb963Tt2jX1hpW3Qq/Pp59+ilGjRnFpKHG/9dZbOH36NK5du9bs\nDFOJZ8+ePZgxYwZ69OjR6HV86623uGMGgJs3byIyMhJeXl5cOkuXLgVjDJ6enjh//jyuX7+OrKws\nJCQkwMLCAuPGjRMSb0ZGBhhjGDJkCCorK8EYw/z5803WuXz5MoKDg41udHz77bdG6XLnJP27cRGZ\nc/HixWCMYenSpQ0++/PPP80yiT///FN4ATcGxhh0Op1w3UWLFoExhsmTJ3PpTJkyxeB68bZQ2rZt\ni9LSUoNjL7/8MnQ6nXq36+joaJLmzJkzMXnyZOTn5zf5NyLSorGeFHNvAquqqjB//nxYWVlh165d\nD/z71NRU7nJTUlKCI0eOwNnZGUQEKysrkzU++eQT9bffbzhBlAnpc/z4cW7DVFrG69evx4wZM9Sb\nkeZkmHfu3EFeXh4KCwuRnZ2Nbdu2qdvAgQNBRMjMzOSOFwD2798PIkJ4eLjZN1DAXy2+pqiqqgJj\nDNu2bTP7HABgZ2cHS0tLAHU9SgkJCWCMYfjw4SZrpaamNtpb5+3tDW9vb8TGxqrHnnvuOaN1heT6\n+q1QXogInp6eBsfefPNNVX/IkCEma44ePRqFhYUGx0SMbTSG0iXLGMPgwYOF67ds2RKMMaSlpXHp\nKF19CryG6ePjY7C/b98+EBG6d++uHnNwcDBJ80E3SGlpaWCM4cMPPzQt2HqIHMNUfrexY7k7d+4U\nZkD6pmkq+oZ5P7QwzISEBG7DDAkJabLlzYsow2yKyspKDBgwQFi6vvfee7C0tIStrS1OnjzJpUVE\nD2yk8BrmzZs3wRjDO++8g6ioKLi5uallPyMjwyzNlJQUdOzYEUuWLFE3hVu3bj06w1TFBGXO4uJi\n+Pv7w8/PDy1btlTHfngq9A0bNiApKUndX7duXQNTNpcJEyYYTPbR6XTw8PDgmvRTUFCADh06qBf1\nhRdeQHh4OJycnMAYQ3h4OK5cucIVt+iKpX///gb7Tz31FNq1a6eOQWRmZuL99983SVP5/VVVVQ0+\n++OPP5CdnY3OnTujrKzM7LibqmTNNUxlnM9Ytm7dCmtra7POVZ933nkHRITFixeb/F0rKysQEQ4e\nPHjfvxNtmGVlZejbt2+D3glzadOmjRqjk5OTkMlVyqQfxth9ezvMRbluIsYagbrhACcnJ3h6emLi\nxIkYPnw4amtrhWjXJz4+HtOnT+fWqd8a1Ol0XOOXjVFYWIiuXbsanMeUSUWaGCbvJKClS5fC09PT\nYFCWN+FqamrQpk0bdYJPdHS0sEJfWVmJjz76CH5+fqphxsbGcmnm5uYiKCjovv3uHTt2NKu1rUBE\n6NWrF9LS0pCWlmbyDNP6fPzxxwb7zs7O8Pf3BwCUlpYiKCgI165dM0lT+a29e/fGihUr1ONpaWno\n378//Pz8zOqy0aexYQWeLlkigp2dndF/P3v2bPTp08esc9XH3ElxQN2NHxFh3rx59/070Yb5xRdf\ncI3F18fa2rrJcVhz0cowFy5ciIULF+KFF14AEcHS0lI9tm/fPiHnuHLlCtq0aYNRo0bhxIkTQjQV\n5s2bB8YY7t27x62lzBxX0nnq1KkCIjTk+eef55o0qYlhmtsS3L59O0aNGtWg4jJn0Lep+D788EOU\nl5cjKChIeLcSUDcuqLQweSkrK8Ps2bPViU6iZ8lu3rwZ1dXVAOpm0jk5OXHF27dvX4N9nU6HxMRE\nAMD8+fOh0+lQUVFhkmZCQgJ0Op36W62srGBlZWVwTEQ3mf5sWf28Z05eNsVQCgsLYWNjwz0GC0Cd\n5GLudezUqROICG+88UaTf3P37l3hhsmbj/VJTk5W4zO19XA/tDDMDRs23HcC19q1a4WcB6gbJiIi\nxMXFCdPcsWMHGGPCJ1QqcxKUukkUubm5BvVm+/btTTblZtEl6+7uDsYYbGxsMG7cOJSUlKCyshK2\ntrbq7EARFBcX47PPPkNBQYHaBcLL/v37DfaPHTumxi2CW7duYfjw4aoxiH4ERkF5nIcHS0tLzJo1\nS93X6XQYO3YsTp8+DWdnZ64x12vXruHtt9/GrVu3cOvWLdy9exdHjhwBYwxnzpzhirsxeLqpX3vt\nNRARVq1a1eTffPnll+jbty+ICEFBQTyhoqqqSm0l83Q/KmOYTRnmrl27DMzIWC5cuKBWUn369MHb\nb7+NyspKAH89bmTstP77odQVnp6euHnzJreePophdu7cWZMySESwsLDA5s2bhWsrhIaGct8UA3XD\nRZGRkRgyZIiQlqU+UVFR8PHxwY0bN4Tqrl27FnZ2dgY33ubQLAxT+REBAQH47rvvMH78eHTp0sWg\nS1Y0K1as4DaI1NRUg3HKs2fPwt/fX+2WFUFMTAwYY1i5cqVmZgmIMczhw4fDw8ND7T5XWoE6na5B\nd60ISktL8fLLL3PPzgPQYBiBxzAvXboELy8vWFtbY+DAgVixYgX279+PuLg4xMXFGTxKMHXqVJSX\nl3PFPnXqVCGtvhMnToCIGkzeqqysxNWrV9VzmPIcZl5eHkJDQ+Hn54elS5ciICAAjDEMGDAAY8aM\ngY2NDeLj47niBuqeH1XiE9mKUtB64QIiErJwyIPOER0dza2jPIu5adMm3Lx5s8GjUkDd/AJTOX36\nNBwcHPDf//6XO0Z9KioqYGFhYWCWTT2X/yCEGSbvuM/Vq1cxZcoUdUWfoKAgJCQk4NatW6JCNCAi\nIoKrgikoKDAwBGVbuHAhsrKyhMR4/vx59a5Wa2JjY4W0uDMzM2Fvbw+dTofJkycjKioKXl5emqzW\noRRcninzwP0X5+Dh8OHDGDFiBAYMGIDnnnsOAwYMwIABAzB37lwcPXqUS1shLy9PbVWdO3eOW8/O\nzu6+3YSmpsnZs2fBGMPixYtRVlaGP/74A1FRUQZdY7NmzcLly5fNjlkZ/yMidOvWzWydpigtLVUn\nimzZskW4PiDOMMeNG2cwBn3v3j28/fbbICKzFw3RZ8eOHfjggw+wfv16dduxYwfee+89REdHw8/P\nD++9955ZPWxeXl5Cx7KBuhXA9B8vmThxYqMTCI1FuGGKXPVHS2bOnMk1yaW6ulp9oFyn02HSpEnc\nj3nUZ+3atXB1dRWu2xiiDPNh0rVrVwwYMECIVmPGIGLlHAUtegfy8vIQGBgIIsKhQ4eEaCrLySld\nhEQELy8v+Pv7IzQ0FJ9++qlJegUFBQgPDwdjDF5eXuosby8vLxw5cgQ//vgjGGNcrUIbGxsQERwc\nHPDDDz+YrXM/IiMj0bdvX+7egKYgMlxZyVxsbW0RGBiI0tJS5OXloX379ur1LC4u5tavrq5GUVFR\nk58rDZyzZ8+arC1yLFvB3t7e4OaM9/px15D6d+eiVvt5GKxYseKRLGZgLDY2NmCMCZ/V1hRffPHF\nA2dHNjc++OADYY8i1F8O73G48VPi1aoX5nFBSQctHvcA6mbYJyYmCl0OsD5EZPRqM39XFFNzcnIS\nNi9BeWbd1Octm8KCBLJ3716Rcppy7do1cnZ2ftRhNEpxcTHdvXuXnJ2dKSAg4KGcc8qUKQ/lPCKZ\nM2eOMK3HKe/qM2zYMGrZsuWjDuORY2NjQ+7u7ppo63Q6evfddzXRlvzFK6+8Qhs2bKDU1FTy9fUV\nohkWFkbJyclERPTxxx9z6zEA4FaRSCQSieRvTrN+H6ZEIpFIJM0FaZgSiUQikRiBNEyJRCKRSIxA\nGqZEIpFIJEYgDVMikUgkEiOQhimRSCQSiRFIw5RIJBKJxAikYUokEolEYgTSMCUSiUQiMQJpmBKJ\nRCKRGIE0TIlEIpFIjEAapkQikUgkRiANUyKRSCQSI5CGKZFIJBKJEUjDlEgkEonECKRhSiQSiURi\nBNIwJRKJRCIxAmmYEolEIpEYgTRMiUQikUiMQBqmRCKRSCRGIA1TIpFIJBIjkIYpkUgkEokRSMOU\nSCQSicQIpGFKJBKJRGIE0jAlEolEIjECaZgSiUQikRiBNEyJRCKRSIxAGqZEIpFIJEYgDVMikUgk\nEiMw2zB/+eUXGjRokMHm6+tLd+7c4Q6qoKCAoqKiqF+/fjRs2DDKyMjg1lTYvXs3hYWF0eDBg2nM\nmDGUk5PDrallWhARbdiwgV566SUaPHgwRUVF0YULF4ToKuzbt498fX3p6tWrQvS0TI/t27dTWFgY\nDRo0SNj1U9Ay31VXV9P8+fPJ19eX8vPzhelqlR5apsWmTZtoyJAhFBwcTNOmTaOqqiohulrmDa2u\nn4LoMkikXcyPY90vLM9BEFu2bMGkSZOEaI0fPx7ffPMNAODQoUOIjY0Vopufn4+uXbvi7NmzAIDv\nv/8eo0ePFqKtj8i0OHfuHLp37478/HwAwJo1axARESFEGwDKy8sxdOhQdO/eHVeuXBGmq4+o9Lh2\n7Rp69OiBq1evAgBWrVqFV155hVtXQat8BwATJkzAwoUL4ePjg7y8PCGaWqaHVmmRnZ2N7t27Izc3\nF7W1tZgyZQr+85//cOtqnTe0uH4KWpVBLWPWp7nX/SLznBDDrKysxMCBA3Hp0iVurdzcXHTr1g1V\nVVUCIjOkqKgI+/fvV/dPnz6Nf/zjH0LPITItAGD79u149dVX1f3z588LjfmTTz7BsmXLEBISoolh\nikyPgoIC/Prrr+p+dnY2unTpwq0LaJvvAOD3338HAKGVl1bpoWVarF692qASzMzMxMsvv8ytq2Xe\nALS5fgpalUEtY1Z4HOp+kXlOyBjm+vXrqUuXLtS+fXturTNnzlDbtm3piy++oNDQUHrttdfo1KlT\nAqIkat26Nb3wwgvq/oEDBygwMFCItoLItCAiCgwMpMuXL1NOTg4BoB07dtA///lPIdrZ2dmUnp5O\n48ePF6LXGCLTw83NjXr37k1ERPfu3aOffvqJXnzxRW5dIm3zHRFR586dhWkpaJUeWqYFY4xqa2vV\n/RYtWtDly5e5dbXMG0TaXD8ibcugVjHr8zjU/ULzHK9719TUoF+/fsjJyeGVAgBs2rQJzz33HHbu\n3AkASElJQUhICKqrq4XoK6Snp+Of//yn2j0rAtFpobBu3To8++yz6Nq1K/r06SPkbq62thajR49G\nRkYGAGjSwtQqPVatWoXu3btj5MiRKCgoEKL5sPKdFnf7otNDy7Q4e/YsOnfujOzsbFRXVyM+Ph7P\nPvsst66CFnlDZTmMIQAAEtFJREFUH5HX72GUQUC7FubjUveLzHPcLczjx49TixYt6JlnnuGVIiIi\nBwcHat26NfXv35+IiEaNGkW3b9+mixcvCtEnItq1axfNnDmTlixZQt7e3sJ0RacFEdGpU6fo66+/\npl27dlFGRgb9+9//ppiYGALApZuSkkLe3t7UtWtXQZE2RIv0ICKKjIykw4cPU2RkJEVERFBlZSW3\n5sPId1ohOj20TAtvb2/64IMPaMqUKfTqq6+St7c3OTg4cOsqaJE3tOJhlEEteVzqfpF5jtsw9+3b\nR3379uWVUXnyySeprKxMbUIzxkin05FOJ+YJmPT0dJo7dy598803FBAQIERTQXRaEBEdOnSIOnfu\nTE8++SQREb300kt07tw5unnzJpfu7t27affu3dS7d2/q3bs35eXl0ciRI+nw4cMiwiYi8enx559/\nUnp6OhHV5YuhQ4dSWVmZkFnDWuc7LdAqPbROi/DwcPrf//5HGzduJB8fH/Lx8eHW1DJvaMXDKINa\n8jjV/aLyHHckZ86coaeffppXRsXX15fc3NwoNTWViIi2bdtGjo6OQvrIKyoqKC4ujpKSkoTGrCA6\nLYiIOnbsSMePH1cNcv/+/eTq6krOzs5cusuXL6dDhw7RwYMH6eDBg+Tp6Unr16+nnj17igibiMSn\nR3FxMU2fPp0KCgqIiOjYsWNUXV1N7dq149bWMt9phVbpoWVaXLp0icLCwqikpISqq6tpyZIlNGLE\nCG5dLfOGVjyMMqglj0vdLzLPWXBFQkT5+fnk4uLCK6PCGKNFixbRzJkzadmyZdS6dWtauHAhWVhw\nh0q7d++m4uJimjp1qsHx77//XshvEJ0WRET9+vWjrKwsioiIICIie3t7SkxMJMaY0PNogej06Nat\nG8XExFBUVBTV1taSlZUVLViwgOzt7bm1tcx3RUVF9Nprr6n748aNoyeeeIJWr15N7u7uZutqlR5a\npkWHDh3oxRdfpLCwMGKM0ZAhQyg8PJxbV8u8odX105KHEfPjUveLzHMMvINhEolEIpH8P6D5DtBI\nJBKJRNKMkIYpkUgkEokRSMOUSCQSicQIpGFKJBKJRGIE0jAlEolEIjECaZgSCRGdPHmSwsPDKTk5\nmXsVpb8bJSUlNGXKFPr5558fdSgSySOl2Rvmzz//TBEREVRTUyNEr7a2loqLi+nIkSM0Y8YM+uCD\nD5r18lmSh0N0dDRt2rSJXn/9dSovL3/U4TQrRo4cSYmJiULfLymR3I/r16/TG2+8Qba2tvThhx8+\n6nD+gmtVW424d+8efvzxRxCRut26dctsvVOnTmHixIkGevpb+/btBUavLXfv3gUArFy5EkOHDsXq\n1atN1igrK2vys127doGIkJuba3aMotm5cycYYwbbmDFjcOPGDSH658+fBxEZvEZNSz755BNhC44X\nFhaiV69eal6OiopCcXGxEG0AsLa2hp2dHU6fPi1MUwsWLlyIOXPmYM6cOYiKisKcOXOQnZ0tRDsk\nJAQWFhawsLAQovcgcnJy1Hy+bds2k79fUFCAlJQUzV5VpyVFRUVgjMHe3h6nTp3CzZs3MWzYMCxa\ntOhRhwZA0PswRXPs2DEDQwsKCuJasT4kJMRAb8KECdizZw/27NkDIoKNjY3A6LUhMTERvXv3hpOT\nEwCgU6dO6u8xlbZt2+LcuXONftavXz/VkO7du8cVswgWLFgAZ2dn6HS6BttXX30l5Bw///wznJ2d\n8csvvwjRexCiDHPPnj3o0qULiAg6nQ5WVlYgIkRGRvIHCeDixYtgjOHEiRNC9PS5e/cuPvvsM9jb\n28PGxgZpaWm4ePEiPv74Y5O1oqKiGr0RFmVwD8Mw9+3bp/5/8ODBah43N0+2bNlSyHtG9ZkxYwbS\n09Nx48YNdcvOzsaMGTOE1RXvvPMOgoKCcObMGfVYRkYG5s2bZ7ZmTU0N5s6dCxsbGxARbG1tzcpn\ngJmGyRiDTqeDv78/unfvrl5cUSgZXtRLmAEgLy8PNTU1TZ7LXPTv7h0dHeHr64sBAwao/168eNEs\n3Rs3biA1NVXVTklJUT+bM2cOiAhDhw7F0aNHTdLVv3ttDOWzfv36oaKiwiTt+q1AZQsKCgJjDKNG\njTJJrymOHz8uzDAvXboEJycnpKWlCYjswcTFxYExxm2YixYtgr29PaKiotRjNTU1sLCwgLW1NW+Y\nWLJkCRhjKC8v59aqz+3btxETE4NDhw4BAA4cOICqqiq0adMGJ0+eNFmvoqICX3/9daOm6eXlxR2v\nlob57rvvgjGG5ORk1NbWIjg4WEh9GhoaKrROBpou34wxnDp1ilu/U6dOsLe3FxDpXwQHBxvEaW1t\nrf5/8eLFJutxGaZOpzP4v4gukKKioofaTWppacllmPPmzVPNsm3btrC0tFQ3IsK0adPM0o2LiwMR\n4ZlnnsEnn3yiHt+/fz+ICA4ODiYbGmC8Ye7fv99k7fsVqPud0xTOnj0LLy8vBAYGCumSzcnJARE9\nNMMkIjDG8N1333HpFBQUNHiX64kTJ0BEQm4kevfujTZt2nDrNEZKSkqDLv/du3dj8+bNZmuWlZUh\nLi4OoaGhwg1Tv9J9/vnnhb1bcvXq1dDpdAgKCgLwl3nqdDokJSVxab///vtCe2GAupZvVFQUXn/9\ndTz33HMG5Zq3cXP9+nUwxhAbGyso2rr3jSrxBQQEIDExEQAwYcIEMMbQv39/3L592yRNk53iq6++\nAmMMhYWFAICsrCxkZWWhVatWyMrKMlWuAS4uLpg4cSK3jrEQESwtLc3+vpWVFUJDQxsc37JlC4gI\n27dvN1nz5MmTICIEBASoxy5evAgHBwcQEa5evWp2vFFRUQ3M67fffsOMGTMQERHBbWx//vknKioq\ncODAARw4cADR0dEGLU0ezp49q96cLV26VMjLv5XuvMYMc/PmzejatSsuX77MfR4AePbZZ0FE8PX1\nFaKnoG8Qf/zxhxBNxhiWL18uREufqVOnonv37gbHYmJihLSGrK2tG7QwlUqSB/0WJmMMkydP5tK7\ne/cu/P39odPpMGvWLAB/pYFiygDUOtZclN6ML7/8kkunMUpLS9VyzTukVVhYCCcnJ6xZs0ZQdHXc\nvXu3yeu1efNmMMbg7e1tkqbJhpmUlAQianAx3dzchNylu7i4IC4ujlvHGJQJLua2ZgsLC2Fpaam+\nMV2fFStWgIjM6mJatWoViAiHDx9Wj/Xs2RNEZPIFro++Yb733nuIjIyEk5OT8JagQkJCgqoZHBxs\ntk58fDy8vLwMxjAdHBzw5ptvcsU3fvz4Rg2ztrYWAQEBICJ07NjR7K51hZMnT4IxBiLCzJkzubQU\nYmJi0LJlSwODWLFiBbdufn4+wsPDDY4dO3YMaWlpJt+R16dNmzYGk85u3LgBxhimTp3KpQuggVkS\nEd5//31u3d9++83AMJ966ikuvZycnAbdrvo9dsq4ZVBQEFdXpzJZztvbG0VFRVwxN4ZSrvv06cOl\ns2jRIjDGhE9S2rp1K5ycnFBQUGBw/Pbt2+jYsaNZdR13C1PB3d0dkyZNMlXOgO3btzcYJ9i8ebM6\nmUHkpN7Lly/D09OTq8/8zp07TRpiYGAgQkJCzNZWfrOtrS1cXV3RoUMHIV1B+oapbEo3obINGTKE\n+zxA3SQBRXP69OlcWjqdDmPHjlX3b9y4gblz50Kn06F9+/bqJKakpCRUVlYarTt58uQGhnnjxg0E\nBATA3d0d06dPh4+PD1feu379upp/7e3tzRo7aYzq6mqcPHkS1dXVqK2tVSfL8d7wPP/886qpu7i4\nqK3iESNGcGn//vvvuH79Oq5cuaLOWreysuKKVZ81a9bA0dERnTt3RqtWrUBE8PDwwLVr17h0S0tL\nER4erpaVjh07mq21b98+ODg4qAa5detWZGVlNTrE5eHhgd9++40rdkVv0KBBXDpA3Xh/y5YtjRp6\nYYzB2dnZKN3PPvtM6E26woEDB8AYw0cffQSgbox/5MiRDeKsra01WtPkWuCnn35C27ZtNTFMoK6A\nKhQVFaFjx44Gd4wiuHr1Kvz9/dUKTAsCAwO5Mml2djZSU1PVmV1EhMDAQHz++edma+bl5aljq/qb\nlZWVwU2J/pgpD76+vmqm3LVrlxDN+pw9exaMMcydOxdt27Y1ecyqsTHM2NhYEBEuXLgAoM5AePLe\nnTt31ILq5eVlMCNSNJ6entzlRDHMH3/8EYwxREdHIzc3V70BMpdTp06pE8CCgoJgY2ODbt26ccVa\nn4yMDBQXF+PNN99U8/Ps2bO5NC9cuCCshbl169YGs731e070DZO39wT4a8y8sWEjc8jIyICfn999\njdLPzw8TJkxAZmamUZrKsI1olC5Zf39/HD58GOPHj28Qa8uWLU3SNKtkZWVlobS01OCYFoapdP8S\nESIiIswe+8nMzESbNm3UyTgWFhaq7p49e7hjbozAwEBs3LiRW8fZ2RlEhLFjx8LJyQm2trb46aef\nTLor0mfdunUGGcbDwwMZGRmIjY1Vj23bto3rMR4FLbp4G0OpYEJCQlSTM5YrV67A1dVVHT8ZM2YM\nLCwsDMZleA0TADZt2iS8S7YxPvroIyGGGRMTAxcXF4ObJ17DBOomEyUkJGDmzJmwsbHhHg9siqNH\nj6plnHfyktaG2dQkShGMHz8eOp0OvXv3FqIH1LXklXLt6OgIDw8PhIWFYceOHUhLSzN57FUZEtKn\ntLQURUVFKCoqauA1xlJTU3NfU2es7mkAUxDWxynKMNu1a4e4uDhUVFSoGX7+/PlITk42qyLIzMxs\ncsECZbOzs+OOuz6iDFPpVtJn5MiRQrunASA3N1eYwSUlJRloafXQ8eHDh9GrVy/uCkZZuMDJycmg\nJ+Po0aMICwsDEeHDDz80W//bb79V08Lc57/0ud8M2/bt23PnjVu3bhnkg5KSEly/fh2DBg2Cra0t\nlzZQNxzi7Oys2eIYp0+fhoeHh3otV65cyaUn0jABoKqqCgEBAQgJCWlgkl5eXkIfpwOA119/Xegj\nJvqGac7CKfVRhkWqqqoMegYGDBig/p93fDMlJcWgPi4sLISLiwuuXLlikk6zM8wlS5aAiDB8+HA1\nsSZNmmT24x/R0dENDNLJyQkrV67EqFGj1GPR0dHcExoUbt26BW9vb7MezahPY13RGzduFG6YCxYs\nUAvBiBEjuLQGDhyozp5btGiRkNZqfdasWYNWrVoZ3Knz4O7ubpBH/Pz84OjoCCJCbGysSeOi9dHv\nwuLlwoULGD16dKOfKc9hisgbSrzDhw9Hhw4d1H1lPMhc7t27B29vb2GPZijor2701ltvqdfRzc2t\nyUU6jOX69etqV7KIMWKgbtZ7WVmZ2h2p5GFRi07ok56erolhenh4NPpsu6n88MMP6pinpaUlkpKS\ncOPGDdTU1KBr165gjAlb1UshPz8f7dq1M/l7zc4wgb+eQdTfQkNDzXqcQt8wdTod1q5d2+BvCgsL\nMXToUGEmNGbMGGFafn5+DbRCQ0OF3OkrHD16VK0Q33rrLW49xTDrt4xFsWzZMvj4+ODUqVMYOHAg\ndDoddu7cya1bXl6OuLg4dTw3Li4Ov//+O7eukrbp6encWhcvXsTQoUMNjhUXF2P8+PGws7NDly5d\n1OUTeTl16pSBSYhoTdjZ2SEhIUFAdIaaTfUepaamCjuP0sK0sLDAqlWrhOn27dvXYAKe6KcE/vjj\nj2bdwgT+eg6TMYaePXuiR48e8Pf3h4WFhSZLVo4ePRp9+/Y1+XvCDNPNzQ2dOnUSonXv3j1s27YN\nbm5uiImJ4cpAV69eVQtP/enF+pw7d07Y2NKrr74qzDD37t0LIlLvyGtqakBEDSpNHpTJHYyx+64z\nayz29vZgjJmVIR/EyZMn4erqigULFqBv377Q6XSIi4vTpBUrCpGGCQCtWrXC1KlT8cMPPyA5OVmd\nxWtlZaXJ4wOiWL16NcaOHSv8WjVlliJvKgHA399fNUxTx77ux88//2wwhln/kR5eHgfDBOoeQRs+\nfLjBTZr+CmciGT16NJydnXH+/HmTvie0hSl6KabHkZKSEu7HEBpjwoQJ6NWrF3r16iV87KeyshLd\nunUTNjlHy4k+69atUyuX4OBgLFiwQJPzNGcSEhIwZswY9OzZE9HR0SZPdHpU+Pv7a6a9aNEi1Sgf\nl/TQ5+bNmwgJCeFe4acxqqqqMH36dCHdp4Bhl+zjyujRo8EYM3meCQPEvPzPw8ODrl+/Luw1XI8r\nhYWF5O7uTkT0//a9is7OznT79m2qra191KFImgn5+fm0Z88e+te//vWoQ5FwUlVVRZ9//jl1796d\n+vfv/6jDMYtvv/2W3n33Xfr111+pU6dORn9PmGFKJBKJRPJ3ptm/QFoikUgkkuaANEyJRCKRSIxA\nGqZEIpFIJEYgDVMikUgkEiOQhimRSCQSiRFIw5RIJBKJxAikYUokEolEYgT/B+neTfEldFiVAAAA\nAElFTkSuQmCC\n","text/plain":["<matplotlib.figure.Figure at 0x7f4a40893d50>"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["======================================================================\n"],"name":"stdout"}]},{"metadata":{"id":"669hzM1OyuWR","colab_type":"code","colab":{"autoexec":{"startup":false,"wait_interval":0},"output_extras":[{"item_id":1}],"base_uri":"https://localhost:8080/","height":375},"outputId":"8e8eb155-1651-4ed3-c397-561d9a95d664","executionInfo":{"status":"ok","timestamp":1521279099842,"user_tz":-540,"elapsed":636,"user":{"displayName":"박경하","photoUrl":"https://lh3.googleusercontent.com/a/default-user=s128","userId":"110612470497084106049"}}},"cell_type":"code","source":["#Avg.cost에 대한 plot이 추가로 필요\n","import pandas as pd\n","\n","%matplotlib inline\n","import matplotlib.pyplot as plt\n","plt.rcParams['figure.figsize']=(6,6) # 기본 크기를 (16,4)로 한다. (단위는 인치)\n","\n","data = {'model0': [2.3082, 2.3033, 2.3037, 2.3030, 2.3035, 2.3031,2.3029,2.3029,2.3028,2.3025,2.3028,2.3025,2.3026,2.3038,2.3027,2.3025,2.3023,2.3027,2.3033,2.3022,2.3021,2.3025,2.3019,2.3022,2.3026,2.3022,2.3017,2.3024,2.3021,2.3021],\n"," 'model3': [0.4326,0.1662,0.1159,0.0882,0.0730,0.0609,0.0511,0.0450,0.0421,0.0367,0.0332,0.0300,0.0295,0.0276,0.0243,0.0247,0.0221,0.0215,0.0214,0.0197,0.0188,0.0195,0.0187,0.0193,0.0177,0.0151,0.0161,0.0155,0.0172,0.0143],\n"," 'model4': [0.4023,0.1467,0.0948,0.0705,0.0526,0.0393,0.0321,0.0259,0.0226,0.0190,0.0179,0.0152,0.0099,0.0130,0.0141,0.0091,0.0101,0.0101,0.0115,0.0092,0.0079,0.0062,0.0088,0.0105,0.0084,0.0054,0.0065,0.0100,0.0054,0.0024]\n"," }\n","df = pd.DataFrame(data)\n","\n","plt.plot(df)\n","plt.show()\n"],"execution_count":49,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAXMAAAFlCAYAAAAZNQgUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n","text/plain":["<matplotlib.figure.Figure at 0x7f4a3d986210>"]},"metadata":{"tags":[]}}]},{"metadata":{"id":"TrEVoG-J7ORf","colab_type":"text"},"cell_type":"markdown","source":["##3.Dropout는 Overfitting을 막는 대표적인 방법이다.\n"," + 방법론들 간의 공통점이 있는가?\n"," + 방법론들 간의 차이점은 무엇인가?\n"," + Overfitting을 고의적으로 유도할 수 있는가?\n"," + Overfitting을 평가하는 방법은 직관적인가?\n","---\n","\n","+ Regulization에서 주의해야 할 점은 무엇인가?\n"," + Regularization Loss 가 Data Loss를 압도할 수 있다.\n"," + 그러므로, 1회차에는 Regularization을 제외한 학습을 통해 data loss를 확인하고, 이후 Regularization을 적용한 학습을 통해 Regularization Loss를 확인한다.\n","+ 방법론들 간의 공통점이 있는가?\n"," + Regulaiztion = Cost function + penalty이다.\n"," + penalty를 통해, 모형을 complexity하게 만들수 있다.\n","+ 방법론들 간의 차이점은 무엇인가?\n"," + L1과 L2는 weights에 penalty를 부여하는 방식이다.\n"," + dropout는 penalty자체를 임의로 부여하는 방식이다.\n","+ Overfitting을 고의적으로 유도할 수 있는가?\n"," + L1 및 L2를 이용하면, 변하는 것을 확인 가능하다.\n"," > ![대체 텍스트](https://i.imgur.com/CJ1vDbz.png)\n"," + dropout를 통해서는 점진적으로 Overfitting을 유발하기 어렵다.\n"," + 이 경우, 모형의 Accuraty가 극단적으로 가는 성향이 있다. (그리고 이것이 binary filter때문인지는 현재는 모르겠다.)"]},{"metadata":{"id":"DX_BppVR7OUf","colab_type":"text"},"cell_type":"markdown","source":["\n","##4.Overfitting은 Learning과정에서 항상 발생한다.\n","\n"," + Dropout은 왜 확률적으로 접근하는가?\n"," + 다른 방법론들도 확률 개념이 들어가는가?\n"," + Dropout은 weigth/bias로 나누어 적용이 가능한가?\n"," ---"]},{"metadata":{"id":"t38hygVU7OY3","colab_type":"text"},"cell_type":"markdown","source":["##5.Dropout의 개념은 이해하기가 어렵다.\n","\n"," + Dropout실행 code를 다른 방식으로 변환할 수 없는가?\n"," + LSTM은 연관이 있는가?\n"," + Dropout로 인해, NN모형=blockbox라고 평가되는 것인가?\n"," + 범위(모형)가 넓은가?\n"]},{"metadata":{"id":"l3F_ZWi6PimN","colab_type":"text"},"cell_type":"markdown","source":["## 8.논문을 접하는 자세\n","+ 기본\n","![대체 텍스트](https://i.imgur.com/Pq7cFr5.jpg)\n","+ 원본\n","![대체 텍스트](https://i.imgur.com/d045J0z.jpg)"]},{"metadata":{"id":"GXcl6qKfwry9","colab_type":"text"},"cell_type":"markdown","source":["## 9.참조목록\n","+ https://deeplearning4j.org/kr/lstm\n","+ http://cs231n.github.io/neural-networks-3/\n","+ http://aikorea.org/cs231n/neural-networks-2-kr/\n","+ Stochastic Dropout: Activation-level Dropout to Learn Better Neural Language Models(https://cs224d.stanford.edu/reports/allen.pdf)\n","+ http://sanghyukchun.github.io/59/\n","+ http://theonly1.tistory.com/329\n"]},{"metadata":{"id":"hJ8cRsjo2E2g","colab_type":"code","colab":{"autoexec":{"startup":false,"wait_interval":0}}},"cell_type":"code","source":[""],"execution_count":0,"outputs":[]}]}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment