한국어 日本語 English 简体中文 繁體中文 हिन्दी Bahasa Indonesia Wikang Tagalog Português Bahasa Melayu ไทย Español Tiếng Việt Deutsch français Türkçe বাংলা Italiano Nederlands русский Українська فارسی ဗမာစာ සිංහල қазақ тілі नेपाली اردو ελληνικά Монгол хэл العربية ខ្មែរ

オーディション情報

ログイン

  import numpy as np
  import matplotlib.pyplot as plt

  # 입력 데이터
  X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
  # 정답 레이블
  y = np.array([[0], [1], [1], [0]])

  # 시그모이드 활성화 함수
  def sigmoid(x):
      return 1 / (1 + np.exp(-x))

  # 시그모이드의 도함수
  def sigmoid_derivative(x):
      return x * (1 - x)

  # 신경망 클래스
  class NeuralNetwork:
      def __init__(self, input_size, hidden_size, output_size):
          # 가중치 초기화
          self.U = np.array([[0.2, 0.2, 0.3], [0.3, 0.1, 0.2]])
          self.W = np.array([[0.3, 0.2], [0.1, 0.4], [0.2, 0.3]])
          # 편향 초기화
          self.b1 = np.zeros((1, hidden_size))
          self.b2 = np.zeros((1, output_size))
          # 에포크와 오차 기록을 위한 리스트
          self.epochs = []
          self.errors = []

      def forward(self, X):
          # 은닉층 계산
          self.hidden_layer = sigmoid(np.dot(X, self.U) + self.b1)
          # 출력층 계산
          self.output_layer = sigmoid(np.dot(self.hidden_layer, self.W) + self.b2)

      def backward(self, X, y, learning_rate):
          # 출력층의 오차 계산
          output_error = y - self.output_layer
          output_delta = output_error * sigmoid_derivative(self.output_layer)

          # 은닉층의 오차 계산
          hidden_error = np.dot(output_delta, self.W.T)
          hidden_delta = hidden_error * sigmoid_derivative(self.hidden_layer)

          # 가중치 및 편향 업데이트
          self.U += learning_rate * np.dot(X.T, hidden_delta)
          self.W += learning_rate * np.dot(self.hidden_layer.T, output_delta)
          self.b1 += learning_rate * np.sum(hidden_delta, axis=0)
          self.b2 += learning_rate * np.sum(output_delta, axis=0)

      def train(self, X, y, epochs, learning_rate):
          for epoch in range(epochs):
              self.forward(X)
              self.backward(X, y, learning_rate)
              self.epochs.append(epoch + 1)
              self.errors.append(np.mean(np.abs(y - self.output_layer)))

      def predict(self, X):
          self.forward(X)
          return self.output_layer

  # 신경망 모델 생성
  input_size = 2
  hidden_size = 3
  output_size = 2
  learning_rate = 1.0
  epochs = 1000

  model = NeuralNetwork(input_size, hidden_size, output_size)
  # 모델 훈련
  model.train(X, y, epochs, learning_rate)

  # 예측 결과 출력
  predictions = model.predict(X)
  for x, y_pred in zip(X, predictions):
      print(f"x1: {x[0]}, x2: {x[1]}, y1: {y_pred[0]:.4f}, y2: {1 - y_pred[0]:.4f}")

  # 에포크와 오차 그래프 출력
  plt.plot(model.epochs, model.errors)
  plt.xlabel('epoch')
  plt.ylabel('Error')
  plt.show()




  def sigmoid(x):
      return 1 / (1 + np.exp(-x))

  def sigmoid_derivative(x):
      return x * (1 - x)

  def mse_loss(y_true, y_pred):
      return ((y_true - y_pred) ** 2).mean()

  def forward_propagation(X, weights, biases):
      layers = [X]
      for weight, bias in zip(weights, biases):
          layers.append(sigmoid(np.dot(layers[-1], weight) + bias))
      return layers

  def back_propagation(y_true, layers, weights, biases, learning_rate=1.0):
      error = y_true - layers[-1]
      for i in reversed(range(len(weights))):
          delta = error * sigmoid_derivative(layers[i+1])
          error = np.dot(delta, weights[i].T)
          weights[i] += learning_rate * np.dot(layers[i].T, delta)
          biases[i] += learning_rate * np.sum(delta, axis=0)
      return weights, biases

  np.random.seed(0)

  weights = [
      np.random.uniform(0, 1, (1, 6)),
      np.random.uniform(0, 1, (6, 4)),
      np.random.uniform(0, 1, (4, 1))
  ]
  biases = [
      np.zeros((1, 6)),
      np.zeros((1, 4)),
      np.zeros((1, 1))
  ]

  df = pd.read_csv('nonlinear.csv')
  X = df['x'].to_numpy().reshape(-1, 1)
  y_true = df['y'].to_numpy().reshape(-1, 1)

  for _ in range(100):
      for i in range(X.shape[0]):
          layers = forward_propagation(X[i:i+1], weights, biases)
          weights, biases = back_propagation(y_true[i:i+1], layers, weights, biases)

  domain = np.linspace(0, 1, 100).reshape(-1, 1)
  y_hat = forward_propagation(domain, weights, biases)[-1]

  plt.scatter(df['x'], df['y'])
  plt.scatter(domain, y_hat, color='r')
  plt.show()


  model = keras.models.Sequential([
      keras.layers.Dense(32, activation='tanh', input_shape=(1,)),
      keras.layers.Dense(16, activation='tanh'),
      keras.layers.Dense(8, activation='tanh'),
      keras.layers.Dense(4, activation='tanh'),
      keras.layers.Dense(1, activation='tanh'),
  ])

  optimizer = keras.optimizers.SGD(learning_rate=0.1)
  model.compile(optimizer=optimizer, loss='mse')

  df = pd.read_csv('nonlinear.csv')
  X = df['x'].to_numpy()
  X = X.reshape(-1, 1)
  y_label = df['y'].to_numpy()

  model.fit(X, y_label, epochs=100)

  domain = np.linspace(0, 1, 100).reshape(-1, 1)
  y_hat = model.predict(domain)
  plt.scatter(df['x'], df['y'])
  plt.scatter(domain, y_hat, color='r')
  plt.show()


  #Iris 데이터를 입력으로 하여 Versicolor, Setosa, Virginica 3종의 품종을 구분하는 심층신경망을 아래 조건을 이용하여 구성
  import numpy as np
  import pandas as pd
  import tensorflow as tf
  from sklearn.model_selection import train_test_split
  from sklearn.preprocessing import StandardScaler
  import matplotlib.pyplot as plt

  # Iris 데이터 불러오기
  url = "https://archive.ics.uci.edu/ml/machine-learning-databases/iris/iris.data"
  df = pd.read_csv(url, header=None)

  # 입력과 출력 데이터 분리
  X = df.iloc[:, :-1].values
  y = df.iloc[:, -1].values

  # 레이블 숫자로 매핑
  label_mapping = {
      'Iris-setosa': 0,
      'Iris-versicolor': 1,
      'Iris-virginica': 2
  }
  y = np.array([label_mapping[label] for label in y])

  # 훈련 데이터와 테스트 데이터로 분할 (stratify 옵션을 사용하여 클래스 비율 유지)
  X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, stratify=y, random_state=42)

  # 특성 스케일링
  scaler = StandardScaler()
  X_train = scaler.fit_transform(X_train)
  X_test = scaler.transform(X_test)

  # 신경망 모델 구성
  model = tf.keras.models.Sequential()
  model.add(tf.keras.layers.Dense(64, activation='relu', input_shape=(4,)))
  model.add(tf.keras.layers.Dropout(0.2))
  model.add(tf.keras.layers.Dense(32, activation='relu'))
  model.add(tf.keras.layers.Dense(10, activation='relu'))
  model.add(tf.keras.layers.Dense(3, activation='softmax'))

  # 모델 컴파일
  model.compile(optimizer='adam',
                loss='sparse_categorical_crossentropy',
                metrics=['accuracy'])

  # 모델 훈련
  history = model.fit(X_train, y_train, batch_size=5, epochs=30, verbose=0)

  # 테스트 데이터에 대한 정확도 평가
  _, test_accuracy = model.evaluate(X_test, y_test)

  # 그래프 출력
  plt.plot(history.history['loss'], label='Loss')
  plt.plot(history.history['accuracy'], label='Accuracy')
  plt.title('Model Loss and Accuracy')
  plt.xlabel('Epoch')
  plt.legend()
  plt.show()

  print("Test Accuracy:", test_accuracy)


  # Fashion MNIST 데이터를 이용하여 다음 조건을 만족하는 신경망을 생성 + 이미지 출력
  import tensorflow as tf
  from tensorflow import keras
  import matplotlib.pyplot as plt
  import numpy as np

  # Fashion MNIST 데이터 불러오기
  fashion_mnist = keras.datasets.fashion_mnist
  (train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()

  # 데이터 전처리: 0~1 사이의 값으로 스케일링
  train_images = train_images / 255.0
  test_images = test_images / 255.0

  # 모델 구성
  model = keras.Sequential([
      keras.layers.Flatten(input_shape=(28, 28)),
      keras.layers.Dense(128, activation='relu'),
      keras.layers.Dropout(0.2),
      keras.layers.Dense(32, activation='relu'),
      keras.layers.Dense(10, activation='softmax')
  ])

  # 모델 컴파일
  model.compile(optimizer='adam',
                loss='sparse_categorical_crossentropy',
                metrics=['accuracy'])

  # 모델 훈련
  history = model.fit(train_images, train_labels, batch_size=64, epochs=10, validation_split=0.25)

  # 테스트 데이터에 대한 정확도 평가
  test_loss, test_accuracy = model.evaluate(test_images, test_labels)

  # 그래프 출력
  plt.figure(figsize=(12, 5))

  # Loss 그래프
  plt.subplot(1, 2, 1)
  plt.plot(history.history['loss'], label='Training Loss')
  plt.plot(history.history['val_loss'], label='Validation Loss', linestyle='dashed')
  plt.title('Model Loss')
  plt.xlabel('Epoch')
  plt.ylabel('Loss')
  plt.legend()

  # Accuracy 그래프
  plt.subplot(1, 2, 2)
  plt.plot(history.history['accuracy'], label='Training Accuracy')
  plt.plot(history.history['val_accuracy'], label='Validation Accuracy', linestyle='dashed')
  plt.title('Model Accuracy')
  plt.xlabel('Epoch')
  plt.ylabel('Accuracy')
  plt.legend()

  plt.tight_layout()
  plt.show()


  # 첫 25개 테스트 이미지 가져오기
  test_images_subset = test_images[:25]
  test_labels_subset = test_labels[:25]

  # 이미지 분류 및 출력
  predictions = model.predict(test_images_subset)
  predicted_labels = np.argmax(predictions, axis=1)

  class_labels = {
      0: "T-shirt/top",
      1: "Trouser",
      2: "Pullover",
      3: "Dress",
      4: "Coat",
      5: "Sandal",
      6: "Shirt",
      7: "Sneaker",
      8: "Bag",
      9: "Ankle boot"
  }

  plt.figure(figsize=(10, 10))
  for i in range(25):
      plt.subplot(5, 5, i + 1)
      plt.imshow(test_images_subset[i])
      plt.title(class_labels[predicted_labels[i]])
      plt.axis('off')
  plt.tight_layout()
  plt.show()

  print("Test Accuracy:", test_accuracy)




  #Keras에서 제공하는 MNIST 데이터에 대하여 합성곱 신경망을 통한 학습을 진행하라. 6만개의 훈련 데이터를 사용하여 학습하고 1만개의 테스트 데이터를 사용하여 정확도를 검증하라
  import tensorflow as tf
  from tensorflow import keras
  from tensorflow.keras import layers

  # 시드 고정
  tf.random.set_seed(0)

  # MNIST 데이터
  (x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()
  x_train = x_train.astype("float32") / 255.0
  x_test = x_test.astype("float32") / 255.0

  # 과제 11-1-1
  # CNN
  model = keras.Sequential()
  model.add(layers.Conv2D(4, (2, 2), strides=(2, 2), padding="same", activation="relu", input_shape=(28, 28, 1)))
  model.add(layers.MaxPooling2D((2, 2)))

  # Flatten
  model.add(layers.Flatten())
  model.add(layers.Dense(32, activation="relu"))
  model.add(layers.Dense(10, activation="softmax"))

  model.compile(
      optimizer=keras.optimizers.Adam(learning_rate=0.001),
      loss=keras.losses.SparseCategoricalCrossentropy(),
      metrics=["accuracy"],
  )

  # 모델 학습
  model.fit(x_train, y_train, epochs=1, batch_size=32)
  test_loss, test_acc = model.evaluate(x_test, y_test)

  print("테스트 데이터의 손실값 : {:.2f}".format(test_loss))
  print("테스트 데이터의 정확도 : {:.2f}".format(test_acc))

  # 과제 11-1-2 98 % 이상되게 하라 정확도가
  # CNN
  model = keras.Sequential()
  model.add(layers.Conv2D(8, (3, 3), strides=(1, 1), padding="same", activation="relu", input_shape=(28, 28, 1)))
  model.add(layers.MaxPooling2D((2, 2)))
  model.add(layers.Conv2D(16, (3, 3), strides=(1, 1), padding="same", activation="relu"))
  model.add(layers.MaxPooling2D((2, 2)))

  # Flatten
  model.add(layers.Flatten())
  model.add(layers.Dense(128, activation="relu"))
  model.add(layers.Dense(10, activation="softmax"))

  model.compile(
      optimizer=keras.optimizers.Adam(learning_rate=0.001),
      loss=keras.losses.SparseCategoricalCrossentropy(),
      metrics=["accuracy"],
  )

  # 모델 학습
  model.fit(x_train, y_train, epochs=1, batch_size=32)
  test_loss, test_acc = model.evaluate(x_test, y_test)

  print("테스트 데이터의 손실값 : {:.2f}, 테스트 데이터의 정확도 : {:.2f}".format(test_loss, test_acc))

  # Keras에서 Fashion MNIST 데이터를 불러와 아래와 같은 신경망을 구성하고 테스트 데이터에 대해 정확도를 계산하라
  import tensorflow as tf
  from tensorflow import keras
  from tensorflow.keras import layers
  import matplotlib.pyplot as plt

  # Fashion MNIST 데이터 로드
  (x_train, y_train), (x_test, y_test) = keras.datasets.fashion_mnist.load_data()
  x_train = x_train.reshape(-1, 28, 28, 1).astype("float32") / 255.0
  x_test = x_test.reshape(-1, 28, 28, 1).astype("float32") / 255.0
  y_train = keras.utils.to_categorical(y_train, num_classes=10)
  y_test = keras.utils.to_categorical(y_test, num_classes=10)

  # 모델 생성
  model = keras.Sequential()
  model.add(layers.Conv2D(32, (3, 3), activation="relu", input_shape=(28, 28, 1)))
  model.add(layers.MaxPooling2D((2, 2)))
  model.add(layers.Conv2D(64, (3, 3), activation="relu"))
  model.add(layers.MaxPooling2D((2, 2)))
  model.add(layers.Conv2D(32, (3, 3), activation="relu"))
  model.add(layers.Flatten())

  # DENSE
  model.add(layers.Dense(1568, activation="relu"))
  model.add(layers.Dense(128, activation="relu"))
  model.add(layers.Dense(32, activation="relu"))

  # 출력층
  model.add(layers.Dense(10, activation="softmax"))

  # 과제 11-2-1
  # 모델 출력
  model.compile(
      optimizer=keras.optimizers.Adam(),
      loss=keras.losses.CategoricalCrossentropy(),
      metrics=["accuracy"],
  )
  model.fit(x_train, y_train, epochs=1)
  _, test_acc = model.evaluate(x_test, y_test)
  print("테스트 정확도: {}".format(test_acc))

  # 과제 11-2-2
  # 테스트 이미지 분류 및 출력
  x_test_subset = x_test[:25]
  y_test_subset = y_test[:25]
  predictions = model.predict(x_test_subset)
  predicted_labels = tf.argmax(predictions, axis=1).numpy()
  class_labels = [
      "T-shirt/top",
      "Trouser",
      "Pullover",
      "Dress",
      "Coat",
      "Sandal",
      "Shirt",
      "Sneaker",
      "Bag",
      "Ankle boot"
  ]

  plt.figure(figsize=(10, 10))
  for i in range(25):
      plt.subplot(5, 5, i + 1)
      plt.imshow(x_test_subset[i].reshape(28, 28))
      plt.title(class_labels[predicted_labels[i]])
      plt.axis('off')
  plt.tight_layout()
  plt.show()



  #51개의 시퀀스를 가지는 Sine 함수 100개를 생성하고, 마지막 시퀀스의 값(51번째 값)을 예측하는 모델을 RNN, LSTM, GRU를 이용하여 구현하고 그래프를 출력
  import numpy as np
  import matplotlib.pyplot as plt
  from tensorflow.keras.models import Sequential
  from tensorflow.keras.layers import SimpleRNN, LSTM, GRU, Dense
  from sklearn.metrics import mean_squared_error

  # 데이터 생성
  x = np.arange(0, 10.2, 0.2)
  num_samples = 100
  num_sequences = 51
  start_points = np.random.choice(100, num_samples)
  dataset = []
  for start in start_points:
      y = np.sin(x + start)
      dataset.append(y)
  dataset = np.array(dataset)
  X = dataset[:, :50]
  y = dataset[:, -1]

  # 모델 생성
  model_list = [SimpleRNN, LSTM, GRU]


  for i, model_name in enumerate(model_list):

      X_train, X_test, y_train, y_test = X[:80], X[80:], y[:80], y[80:]
      model = Sequential()
      model.add(model_name(10, input_shape=(50, 1)))
      model.add(Dense(1))
      model.compile(optimizer='adam', loss='mean_squared_error')

      # 모델 학습
      history = model.fit(X_train[:, :, np.newaxis], y_train, epochs=50, verbose=0)
      y_train_pred = model.predict(X_train[:, :, np.newaxis])
      y_test_pred = model.predict(X_test[:, :, np.newaxis])
      train_mse = mean_squared_error(y_train, y_train_pred)
      test_mse = mean_squared_error(y_test, y_test_pred)

      # 출력
      fig, axs = plt.subplots(2, 2, figsize=(10, 8))
      axs[0, 0].plot(history.history['loss'])
      axs[0, 0].set_xlabel('epoch')
      axs[0, 0].set_ylabel('loss')

      axs[1, 0].plot(y_train, label='Train Actual', color='black', linewidth=2.0)
      axs[1, 0].plot(y_train_pred, label='Train Predicted', color='red')
      axs[1, 0].set_title('Train')
      axs[1, 0].legend(loc='upper right')

      axs[0, 1].scatter(y_train, y_train_pred, label='Train', marker='o')
      axs[0, 1].scatter(y_test, y_test_pred, label='Test', color='orange', marker='o')
      axs[0, 1].set_xlabel('y')
      axs[0, 1].set_ylabel('y_hat')
      axs[0, 1].legend(loc='upper left')

      axs[1, 1].plot(y_test, label='Test Actual', color='black', linewidth=2.0)
      axs[1, 1].plot(y_test_pred, label='Test Predicted', color='red')
      axs[1, 1].set_title('Test')
      axs[1, 1].legend(loc='upper right')

      axs[1, 0].text(0.05, 0.9, f'MSE: {train_mse:.5f}', transform=axs[1, 0].transAxes)
      axs[1, 1].text(0.05, 0.9, f'MSE: {train_mse:.5f}', transform=axs[1, 1].transAxes)

      plt.tight_layout()
      plt.show()