LSTM_network.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "LSTM_network.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyPOk2ys5NUfwR0zrNu/+Gdb",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/alessio76/Progetto-ML/blob/main/LSTM_network.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "62rOxKU40my5"
      },
      "source": [
        "import numpy as np\n",
        "import keras\n",
        "import tensorflow as tf\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n",
        "from matplotlib import pyplot\n",
        "import json\n",
        "from google.colab import files\n",
        "import copy\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "import time\n",
        "from sklearn.datasets import load_breast_cancer"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "83RJK8sMMqRM"
      },
      "source": [
        "def build_model_n_layer(activation_hidden='relu', recurrent_activation='sigmoid', learning_rate=0.1, momentum=0, optimizer='SGD', n_neurons = 5, \n",
        "                loss = 'sparse_categorical_crossentropy', regularization = None, recurrent_regularization=None, n_layers = 1, n_features = 30, init_mode = 'glorot_uniform'):\n",
        "   n_classes = 2\n",
        "   hidden_layers = list()\n",
        "\n",
        "   for i in range(n_layers-1):\n",
        "     layer = tf.keras.layers.LSTM(n_neurons, activation=activation_hidden, recurrent_activation=recurrent_activation, kernel_regularizer= regularization,\n",
        "                                  recurrent_regularizer=recurrent_regularization, return_sequences=True, kernel_initializer=init_mode)\n",
        "     hidden_layers.append(layer)\n",
        "     \n",
        "   first_layer = [tf.keras.layers.LSTM(n_features, activation=activation_hidden, input_shape=(n_features,1), recurrent_activation=recurrent_activation, \n",
        "                                       kernel_regularizer= regularization, recurrent_regularizer=recurrent_regularization, return_sequences = True, kernel_initializer=init_mode)]\n",
        "\n",
        "   last_layer = [tf.keras.layers.LSTM(n_neurons, activation=activation_hidden, recurrent_activation=recurrent_activation, kernel_regularizer= regularization,\n",
        "                                      recurrent_regularizer=recurrent_regularization, kernel_initializer=init_mode),\n",
        "                tf.keras.layers.Dense(n_classes, activation='softmax', kernel_initializer=init_mode)]            \n",
        "   \n",
        "   total_layers = first_layer + hidden_layers + last_layer                          \n",
        "   model = tf.keras.Sequential(total_layers)\n",
        "  \n",
        "   if optimizer == 'SGD':\n",
        "         optimizer=tf.keras.optimizers.SGD(learning_rate=learning_rate, momentum=momentum)\n",
        "\n",
        "\n",
        "   model.compile(loss=loss, optimizer=optimizer, metrics = 'accuracy')\n",
        "   return model"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "K12ZF0AWY7Z8"
      },
      "source": [
        "def scoring(train_x, train_y, test_x, test_y, model):\n",
        "  y_pred_test, y_pred_train = model.predict(input_test), model.predict(input_train)\n",
        "  y_pred_test, y_pred_train = transform_y_pred(y_pred_test), transform_y_pred(y_pred_train)\n",
        "  compute_precision, compute_recall = tf.keras.metrics.Precision(), tf.keras.metrics.Recall()\n",
        "  compute_precision.update_state(train_y, y_pred_train)\n",
        "  compute_recall.update_state(train_y, y_pred_train)\n",
        "  train_precision, train_recall = compute_precision.result(), compute_recall.result()\n",
        "\n",
        "  compute_precision.update_state(test_y, y_pred_test)\n",
        "  compute_recall.update_state(test_y, y_pred_test)\n",
        "  test_precision, test_recall = compute_precision.result(), compute_recall.result()\n",
        "\n",
        "  print(\"Train:\", \"Precision = \",train_precision, \"Recall = \", train_recall, '\\n')\n",
        "  print(\"Test:\", \"Precision = \",test_precision, \"Recall = \", test_recall, '\\n')"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qt6UFFS7ciKo"
      },
      "source": [
        "def transform_y_pred(pred_y):\n",
        "  transformed_y = list()\n",
        "  for i in range(pred_y.shape[0]):\n",
        "    actual_max = max(pred_y[i, :])\n",
        "    transformed_y.append(list(np.where(pred_y[i,:] == actual_max))[0])\n",
        "  return transformed_y\n"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ithbbijENjY4"
      },
      "source": [
        "def transform_x(x):  \n",
        "  length = x.shape[0]\n",
        "  transformed_x = np.array(x)\n",
        "  transformed_x = np.reshape(transformed_x, (transformed_x.shape[0], transformed_x.shape[1], 1))\n",
        "  return transformed_x"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sxzFwI28skjN"
      },
      "source": [
        "def plot(history):\n",
        " #plotter = tfdocs.plots.HistoryPlotter(smoothing_std=0)\n",
        " epochs = history.epoch\n",
        " plt.plot(epochs, history.history['accuracy'])\n",
        " plt.xlabel(\"epochs\")\n",
        " plt.ylabel('Accuracy_train')\n",
        " plt.show()\n",
        "\n",
        " plt.plot(epochs, history.history['val_accuracy'])\n",
        " plt.xlabel(\"epochs\")\n",
        " plt.ylabel('Accuracy_val')\n",
        " plt.show()\n",
        "\n",
        " plt.plot(epochs, history.history['loss'])\n",
        " plt.xlabel(\"epochs\")\n",
        " plt.ylabel('Loss_train')\n",
        " plt.show()\n",
        "\n",
        " plt.plot(epochs, history.history['val_loss'])\n",
        " plt.xlabel(\"epochs\")\n",
        " plt.ylabel('Loss_test')\n",
        " plt.show()"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VM_QEWWyVxba",
        "outputId": "62711455-8f3a-4029-a959-4f6735b69b17"
      },
      "source": [
        "#dataset è un dizionario\n",
        "dataset = load_breast_cancer()\n",
        "x,y = dataset[\"data\"], dataset[\"target\"]\n",
        "train_x, test_x, train_y, test_y = train_test_split(x, y, test_size=.2, random_state=1)\n",
        "print(train_x.shape, test_x.shape)\n",
        "scaler = StandardScaler()\n",
        "scaler.fit(train_x)\n",
        "train_x = scaler.transform(train_x)\n",
        "test_x = scaler.transform(test_x)"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(455, 30) (114, 30)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tiYddWCWuiWq"
      },
      "source": [
        "Anche qui ADAM più regolare, L1 migliora regolarità più di L2"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Lmx4QLA1bxRK",
        "outputId": "154b7b0a-320c-44eb-a314-5505689cf15f"
      },
      "source": [
        "print(model_n_layers_param)"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'n_neurons': 10, 'n_layers': 4, 'activation_hidden': 'tanh', 'learning_rate': 0.3, 'regularization': <keras.regularizers.L1 object at 0x7fc855dc8d90>, 'recurrent_regularization': <keras.regularizers.L1 object at 0x7fc855dc8b50>}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lRmDReXBeP_v"
      },
      "source": [
        "soltanto glorot e he danno performance accettabili, glorot leggermente superiore. L2 meglio di L1, SGD non è adatto"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "08R_WmPEWxYA",
        "outputId": "06096207-e9eb-4187-a751-cef7dfcfde82"
      },
      "source": [
        "model_n_layers_param = dict()\n",
        "model_n_layers_param[\"n_neurons\"] = 20\n",
        "model_n_layers_param[\"n_layers\"] = 5\n",
        "model_n_layers_param[\"activation_hidden\"] = 'tanh'\n",
        "model_n_layers_param[\"optimizer\"] = 'Adam'\n",
        "model_n_layers_param[\"learning_rate\"] = .3\n",
        "model_n_layers_param[\"regularization\"] = tf.keras.regularizers.L2(.01)\n",
        "model_n_layers_param[\"recurrent_regularization\"] = tf.keras.regularizers.L2(.001)\n",
        "init_mode = ['random_uniform', 'random_normal', 'zero', 'glorot_normal', 'glorot_uniform','he_normal', 'he_uniform']\n",
        "model_n_layers_param[\"init_mode\"] = 'glorot_uniform'\n",
        "batch_size = 50\n",
        "epochs = 12\n",
        "model4 = build_model_n_layer(**model_n_layers_param)\n",
        "model4.summary()\n",
        "early_stop = tf.keras.callbacks.EarlyStopping(monitor='loss', patience=20, restore_best_weights=True)\n",
        "start = time.time()\n",
        "input_train, input_test = transform_x(train_x), transform_x(test_x)\n",
        "history = model4.fit(input_train, train_y, epochs=epochs, batch_size=batch_size,  validation_split = 0.1, shuffle=True, verbose=0, callbacks=[early_stop])\n",
        "print('\\nTraining MLP completato in %.2f s.' % (time.time() - start))\n",
        "scoring(input_train, train_y, input_test, test_y, model4)\n",
        "plot(history)"
      ],
      "execution_count": 81,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_61\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm_345 (LSTM)              (None, 30, 30)            3840      \n",
            "_________________________________________________________________\n",
            "lstm_341 (LSTM)              (None, 30, 20)            4080      \n",
            "_________________________________________________________________\n",
            "lstm_342 (LSTM)              (None, 30, 20)            3280      \n",
            "_________________________________________________________________\n",
            "lstm_343 (LSTM)              (None, 30, 20)            3280      \n",
            "_________________________________________________________________\n",
            "lstm_344 (LSTM)              (None, 30, 20)            3280      \n",
            "_________________________________________________________________\n",
            "lstm_346 (LSTM)              (None, 20)                3280      \n",
            "_________________________________________________________________\n",
            "dense_61 (Dense)             (None, 2)                 42        \n",
            "=================================================================\n",
            "Total params: 21,082\n",
            "Trainable params: 21,082\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "\n",
            "Training MLP completato in 28.05 s.\n",
            "Train: Precision =  tf.Tensor(0.9641577, shape=(), dtype=float32) Recall =  tf.Tensor(0.94385964, shape=(), dtype=float32) \n",
            "\n",
            "Test: Precision =  tf.Tensor(0.95988536, shape=(), dtype=float32) Recall =  tf.Tensor(0.93837535, shape=(), dtype=float32) \n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cFnXerQ1XjhC"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}