ML_LAB5_21171.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "# RICHA VIVEK SAVANT - BL.EN.U4CSE21171"
      ],
      "metadata": {
        "id": "X05yN_NII7AZ"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "aejna_xVdmpe"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "from sklearn.neural_network import MLPClassifier\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.linear_model import Perceptron"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "kUpZNqUfdmpn",
        "outputId": "48cbd975-d7fb-4047-e02f-aaa34109c2d3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 507
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of epochs needed for convergence: 130\n",
            "Final weights: W0 = -0.10000000000000765, W1 = 0.1000000000000001, W2 = 0.05000000000000032\n"
          ]
        }
      ],
      "source": [
        "'''\n",
        "A1.Develop the above perceptron in your own code (don’t use the perceptron model available from package). Use the initial weights as provided below. W0 = 10, W1 = 0.2, w2 = -0.75, learning rate (α) = 0.05Use Stepactivation function to learn the weights of the network to implement above provided AND gate logic. The activation function is demonstratedbelow.Identify the number of epochs needed for the weights to converge in the learning process. Make a plot of the epochs against the error values calculated (after each epoch, calculate the sum-square-error against all training samples). (Note: Learning is said to be converged if the error is less than or equal to 0.002. Stop the learning after 1000 iterations if the convergence error condition is not met)\n",
        "'''\n",
        "\n",
        "def train_perceptron_for_and_gate(initialWeights, learningRate=0.05, convergenceError=0.002, maxEpochs=1000):\n",
        "    # Initialize weights and bias\n",
        "    W0, W1, W2 = initialWeights\n",
        "\n",
        "    # Step activation function\n",
        "    def stepActivation(x):\n",
        "        return 1 if x >= 0 else 0\n",
        "\n",
        "    # Perceptron output calculation\n",
        "    def perceptronOutput(x1, x2):\n",
        "        return W0 + W1 * x1 + W2 * x2\n",
        "\n",
        "    # AND gate training data\n",
        "    trainingData = [\n",
        "        (0, 0, 0),\n",
        "        (0, 1, 0),\n",
        "        (1, 0, 0),\n",
        "        (1, 1, 1)\n",
        "    ]\n",
        "\n",
        "    # Training loop\n",
        "    epochs = 0\n",
        "    errors = []\n",
        "\n",
        "    while True:\n",
        "        epoch_error = 0\n",
        "        for x1, x2, target in trainingData:\n",
        "            output = perceptronOutput(x1, x2)\n",
        "            error = target - stepActivation(output)\n",
        "            epoch_error = epoch_error + error ** 2\n",
        "            W0 = W0 + learningRate * error\n",
        "            W1 = W1 + learningRate * error * x1\n",
        "            W2 = W2 + learningRate * error * x2\n",
        "\n",
        "        errors.append(epoch_error)\n",
        "        epochs += 1\n",
        "\n",
        "        if epoch_error <= convergenceError or epochs >= maxEpochs:\n",
        "            break\n",
        "\n",
        "    return epochs, [W0, W1, W2], errors\n",
        "\n",
        "# Initial weights\n",
        "initial_weights = [10, 0.2, -0.75]\n",
        "\n",
        "# Train the perceptron for the AND gate\n",
        "num_epochs, final_weights, error_history = train_perceptron_for_and_gate(initial_weights)\n",
        "\n",
        "# Plotting epochs against error values\n",
        "plt.plot(range(num_epochs), error_history)\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Error')\n",
        "plt.title('Error vs. Epochs')\n",
        "plt.show()\n",
        "\n",
        "print(f\"Number of epochs needed for convergence: {num_epochs}\")\n",
        "print(f\"Final weights: W0 = {final_weights[0]}, W1 = {final_weights[1]}, W2 = {final_weights[2]}\")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ji09PdJudmpq"
      },
      "source": [
        "A2. Repeat the above A1 experiment with following activation functions. Compare the iterations taken to converge against each of the activation functions. Keep the learning rate same as A1."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "id": "Olh61NBXdmpr",
        "outputId": "9e0ce03b-2f24-46d9-cbba-69d64213a2fb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 772
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x600 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Activation Function: Step\n",
            "Number of Epochs: 130\n",
            "Final Weights: W0 = -0.10000000000000765, W1 = 0.1000000000000001, W2 = 0.05000000000000032\n",
            "\n",
            "Activation Function: Sigmoid\n",
            "Number of Epochs: 1000\n",
            "Final Weights: W0 = -6.1353130787344305, W1 = 3.970833206406072, W2 = 3.963948347259991\n",
            "\n",
            "Activation Function: ReLU\n",
            "Number of Epochs: 390\n",
            "Final Weights: W0 = -0.9277447198109581, W1 = 0.9511600082912268, W2 = 0.9498108559260638\n",
            "\n"
          ]
        }
      ],
      "source": [
        "def step_activation(x):\n",
        "    return 1 if x >= 0 else 0\n",
        "\n",
        "def sigmoid_activation(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "def relu_activation(x):\n",
        "    return max(0, x)\n",
        "\n",
        "def train_perceptron(initial_weights, learningRate=0.05, activation_func=step_activation, convergence_error=0.002, max_epochs=1000):\n",
        "    W0, W1, W2 = initial_weights\n",
        "\n",
        "    def perceptron_output(x1, x2):\n",
        "        return W0 + W1 * x1 + W2 * x2\n",
        "\n",
        "    training_data = [\n",
        "        (0, 0, 0),\n",
        "        (0, 1, 0),\n",
        "        (1, 0, 0),\n",
        "        (1, 1, 1)\n",
        "    ]\n",
        "\n",
        "    epochs = 0\n",
        "    errors = []\n",
        "\n",
        "    while True:\n",
        "        epoch_error = 0\n",
        "        for x1, x2, target in training_data:\n",
        "            output = perceptron_output(x1, x2)\n",
        "            error = target - activation_func(output)\n",
        "            epoch_error = epoch_error + error ** 2\n",
        "            W0 = W0 + learningRate * error\n",
        "            W1 = W1 + learningRate * error * x1\n",
        "            W2 = W2 + learningRate * error * x2\n",
        "        errors.append(epoch_error)\n",
        "        epochs += 1\n",
        "\n",
        "        if epoch_error <= convergence_error or epochs >= max_epochs:\n",
        "            break\n",
        "\n",
        "    return epochs, [W0, W1, W2], errors\n",
        "\n",
        "# Initial weights\n",
        "initial_weights = [10, 0.2, -0.75]\n",
        "\n",
        "# Train the perceptron with different activation functions\n",
        "activation_functions = [step_activation, sigmoid_activation, relu_activation]\n",
        "activation_names = [\"Step\", \"Sigmoid\", \"ReLU\"]\n",
        "\n",
        "num_epochs_list = []\n",
        "final_weights_list = []\n",
        "for activation_func in activation_functions:\n",
        "    num_epochs, final_weights, _ = train_perceptron(initial_weights, activation_func=activation_func)\n",
        "    num_epochs_list.append(num_epochs)\n",
        "    final_weights_list.append(final_weights)\n",
        "\n",
        "# Plotting epochs vs. activation functions\n",
        "plt.figure(figsize=(10, 6))\n",
        "plt.bar(activation_names, num_epochs_list, color=['pink', 'lavender', 'purple'])\n",
        "plt.xlabel('Activation Function')\n",
        "plt.ylabel('Number of Epochs')\n",
        "plt.title('Comparison of Convergence (Epochs) for Different Activation Functions')\n",
        "plt.show()\n",
        "\n",
        "# Print the number of epochs and final weights for each activation function\n",
        "for i, activation_name in enumerate(activation_names):\n",
        "    print(f\"Activation Function: {activation_name}\")\n",
        "    print(f\"Number of Epochs: {num_epochs_list[i]}\")\n",
        "    print(f\"Final Weights: W0 = {final_weights_list[i][0]}, W1 = {final_weights_list[i][1]}, W2 = {final_weights_list[i][2]}\\n\")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9QBHRvKYdmps"
      },
      "source": [
        "A3. Repeat exercise A1 with varying the learning rate, keeping the initial weights same. Take learning rate = {0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1}. Make a plot of the number of iterations taken for learning to converge against the learning rates."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "4mpFjK7ddmpu",
        "outputId": "187d964c-62c2-4917-e856-2d64506dede6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "def step_activation(x):\n",
        "    return 1 if x >= 0 else 0\n",
        "\n",
        "def train_perceptron(initial_weights, learningRate=0.05, activation_func=step_activation, convergence_error=0.002, max_epochs=1000):\n",
        "    W0, W1, W2 = initial_weights\n",
        "\n",
        "    def perceptron_output(x1, x2):\n",
        "        return W0 + W1 * x1 + W2 * x2\n",
        "\n",
        "    #and gate logic\n",
        "    training_data = [\n",
        "        (0, 0, 0),\n",
        "        (0, 1, 0),\n",
        "        (1, 0, 0),\n",
        "        (1, 1, 1)\n",
        "    ]\n",
        "\n",
        "    epochs = 0\n",
        "    errors = []\n",
        "\n",
        "    while True:\n",
        "        epoch_error = 0\n",
        "        for x1, x2, target in training_data:\n",
        "            output = perceptron_output(x1, x2)\n",
        "            error = target - activation_func(output)\n",
        "            epoch_error = epoch_error + error ** 2\n",
        "            W0 = W0 + learningRate * error\n",
        "            W1 = W1 + learningRate * error * x1\n",
        "            W2 = W2 + learningRate * error * x2\n",
        "\n",
        "        errors.append(epoch_error)\n",
        "        epochs += 1\n",
        "\n",
        "        if epoch_error <= convergence_error or epochs >= max_epochs:\n",
        "            break\n",
        "\n",
        "    return epochs, [W0, W1, W2], errors\n",
        "\n",
        "# Initial weights\n",
        "initial_weights = [10, 0.2, -0.75]\n",
        "\n",
        "# List of learning rates to test\n",
        "learning_rates = [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1]\n",
        "\n",
        "# List to store the number of epochs for each learning rate\n",
        "num_epochs_list = []\n",
        "\n",
        "# Train the perceptron with varying learning rates\n",
        "for learning_rate in learning_rates:\n",
        "    num_epochs, final_weights, errors = train_perceptron(initial_weights, learningRate=learning_rate)\n",
        "    num_epochs_list.append(num_epochs)\n",
        "\n",
        "# Plotting learning rates vs. number of epochs\n",
        "plt.plot(learning_rates, num_epochs_list, marker='o', linestyle='-')\n",
        "plt.xlabel('Learning Rate')\n",
        "plt.ylabel('Number of Epochs')\n",
        "plt.title('Number of Epochs vs. Learning Rate')\n",
        "plt.grid(True)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4DD--Orydmpv"
      },
      "source": [
        "A4. Repeat the above exercises, A1 to A3,for XOR gate logic"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k2a-TmiIdmpw"
      },
      "source": [
        "A1 using XOR gate logic"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "6BaenZGvdmpx",
        "outputId": "926c9778-2762-42a9-e921-ad191e60f136",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 507
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of epochs needed for convergence: 1000\n",
            "Final weights: W0 = 0.09999999999999236, W1 = -0.09999999999999969, W2 = -0.09999999999999969\n"
          ]
        }
      ],
      "source": [
        "def train_perceptron_for_and_gate(initialWeights, learningRate=0.05, convergenceError=0.002, maxEpochs=1000):\n",
        "    # Initialize weights and bias\n",
        "    W0, W1, W2 = initialWeights\n",
        "\n",
        "    # Step activation function\n",
        "    def stepActivation(x):\n",
        "        return 1 if x >= 0 else 0\n",
        "\n",
        "    # Perceptron output calculation\n",
        "    def perceptronOutput(x1, x2):\n",
        "        return W0 + W1 * x1 + W2 * x2\n",
        "\n",
        "    # XOR gate training data\n",
        "    trainingData = [\n",
        "        (0, 0, 0),\n",
        "        (0, 1, 1),\n",
        "        (1, 0, 1),\n",
        "        (1, 1, 0)\n",
        "    ]\n",
        "\n",
        "    # Training loop\n",
        "    epochs = 0\n",
        "    errors = []\n",
        "\n",
        "    while True:\n",
        "        epoch_error = 0\n",
        "        for x1, x2, target in trainingData:\n",
        "            output = perceptronOutput(x1, x2)\n",
        "            error = target - stepActivation(output)\n",
        "            epoch_error = epoch_error + error ** 2\n",
        "            W0 = W0 + learningRate * error\n",
        "            W1 = W1 + learningRate * error * x1\n",
        "            W2 = W2 + learningRate * error * x2\n",
        "\n",
        "        errors.append(epoch_error)\n",
        "        epochs += 1\n",
        "\n",
        "        if epoch_error <= convergenceError or epochs >= maxEpochs:\n",
        "            break\n",
        "\n",
        "    return epochs, [W0, W1, W2], errors\n",
        "\n",
        "# Initial weights\n",
        "initial_weights = [10, 0.2, -0.75]\n",
        "\n",
        "# Train the perceptron for the AND gate\n",
        "num_epochs, final_weights, error_history = train_perceptron_for_and_gate(initial_weights)\n",
        "\n",
        "# Plotting epochs against error values\n",
        "plt.plot(range(num_epochs), error_history)\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Error')\n",
        "plt.title('Error vs. Epochs')\n",
        "plt.show()\n",
        "\n",
        "print(f\"Number of epochs needed for convergence: {num_epochs}\")\n",
        "print(f\"Final weights: W0 = {final_weights[0]}, W1 = {final_weights[1]}, W2 = {final_weights[2]}\")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DRUQR8-vdmp0"
      },
      "source": [
        "A2 using XOR gate logic"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "id": "RJDkmuNPdmp5",
        "outputId": "f70e0ad1-62c1-4e85-aaa9-2f18ca9ea833",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 524
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjcAAAHHCAYAAABDUnkqAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABR7UlEQVR4nO3deVxUVeMG8OcOAwPIprKKbC4vrrimoiaZGi6VmEsuveKepqlZampq2mtYZmppbpVaaZSWZq6RiuaaGypqlOb2UwGXYFgEgTm/P4zLjCDrheHC8/185qNz75kzZy7UPJ7lHkkIIUBERERUQWjM3QAiIiIiJTHcEBERUYXCcENEREQVCsMNERERVSgMN0RERFShMNwQERFRhcJwQ0RERBUKww0RERFVKAw3REREVKEw3BCRqrz77ruQJAl3794ttffIzMzElClT4OXlBY1Gg5CQkFJ7LyJSHsMNVQpr166FJElPfBw9etTcTSw3ssPDkx6xsbHmbmKp+/LLL7FgwQL06dMH69atwxtvvFFq7xUfH49q1arh2WefzXUuIyMDjRs3hq+vL1JSUuTj58+fxyuvvAJPT0/odDrUqFEDgwYNwvnz53PV8fjvvlarhaenJ4YMGYKbN28Wqa1nz57F0KFD4efnB2tra9jZ2aFp06aYMmUK/v7776J/eAA7duzAu+++W6zXEj2J1twNICpLc+fOhZ+fX67jderUMUNryrfly5fDzs4u13EnJ6eyb0wZ27t3Lzw9PbFo0aJSfy9XV1d88MEHGDVqFNatW4fQ0FD53MKFCxEdHY2ff/4ZVapUAQD8+OOPGDBgAKpVq4bhw4fDz88PV69exRdffIFNmzYhPDwcvXr1yvU+2b/7aWlpOHr0KNauXYuDBw8iOjoa1tbWBbZz9erVGDNmDJydnTFo0CDUq1cPmZmZiI6OxldffYXFixfjwYMHsLCwKNLn37FjB5YtW8aAQ8oSRJXAmjVrBABx/PjxIr82IyNDpKen53kuOTm5RO0yGAwiNTW1RHUobfbs2QKAuHPnjrmbkqeyaF/Hjh1Fw4YNFasvKytLPHjw4InnDQaDaN++vXB2dhZ3794VQgjx999/CxsbG/HSSy/J5S5duiRsbW1FvXr1RHx8vEkdd+7cEfXq1RNVqlQRly9flo8/6Xd/6tSpAoD47rvvCmz/oUOHhIWFhejQoYPQ6/W5zj948EC88847IjMzs8C6Hjd27FjBryJSGoeliIxcvXoVkiTho48+wuLFi1G7dm3odDpcuHBBHq65cOECBg4ciKpVq6J9+/YAHs3ReO+99+Tyvr6+mD59OtLT003q9/X1xfPPP4/du3ejZcuWsLGxwcqVK/Nsy7hx42BnZ4fU1NRc5wYMGAB3d3dkZWUBAE6cOIHg4GA4OzvDxsYGfn5+GDZsmMJXx1RkZCQkScJ3332H6dOnw93dHVWqVMGLL76IGzdu5Cq/ceNGtGjRAjY2NnB2dsYrr7yS57DIH3/8gX79+sHFxQU2Njbw9/fHjBkzcpVLSEjAkCFD4OTkBEdHRwwdOjTXtYqIiED79u3h5OQEOzs7+Pv7Y/r06U/8TNk//3379uH8+fPyUE5kZCQAICUlBW+++Sa8vLyg0+ng7++Pjz76CEIIk3okScK4ceOwfv16NGzYEDqdDrt27Xri+0qShBUrViAxMRFvvfUWAOC1116DVqvFJ598IpdbsGABUlNTsWrVKri4uJjU4ezsjJUrVyIlJQUffvjhE98r29NPPw0AuHz5coFl58yZA0mSsH79etjb2+c6b21tjffee8+k1+a3335D37594e3tDZ1OBy8vL7zxxht48OCBXGbIkCFYtmyZfA2yH9kMBgMWL16Mhg0bwtraGm5ubnj11Vfxzz//FNhmqtw4LEWVSmJiYq6JqJIkoXr16ibH1qxZg7S0NIwaNQo6nQ7VqlWTz/Xt2xd169bF+++/L3+pjRgxAuvWrUOfPn3w5ptv4tixYwgLC8PFixexefNmk7pjYmIwYMAAvPrqqxg5ciT8/f3zbOvLL7+MZcuWYfv27ejbt698PDU1FT///DOGDBkCCwsLxMfH47nnnoOLiwvefvttODk54erVq/jxxx9LdK3u37+f65hWq801LDVv3jxIkoSpU6ciPj4eixcvRufOnREVFQUbGxsAj+Z9DB06FE899RTCwsIQFxeHJUuW4NChQzh9+rRc59mzZ/H000/D0tISo0aNgq+vLy5fvoyff/4Z8+bNM3nffv36wc/PD2FhYTh16hQ+//xzeYgHeDQv5fnnn0dAQADmzp0LnU6HS5cu4dChQ0/8zC4uLvj6668xb948JCcnIywsDABQv359CCHw4osvYt++fRg+fDiaNm2K3bt3Y/Lkybh582auIay9e/fi+++/x7hx4+Ds7AxfX998r3fDhg3x1ltvISwsDPb29ti1axeWLFkCT09PuczPP/8MX19fOZg8rkOHDvD19cX27dvzfS/gUZADgKpVq+ZbLjU1FXv37sUzzzyDmjVrFlhvto0bNyI1NRVjxoxB9erV8fvvv+PTTz/F//3f/2Hjxo0AgFdffRW3bt1CREQEvv7661x1vPrqq/Lvzvjx43HlyhUsXboUp0+fxqFDh2BpaVno9lAlY+aeI6Iykd01n9dDp9PJ5a5cuSIACAcHh1zd/tnDIQMGDDA5HhUVJQCIESNGmBx/6623BACxd+9e+ZiPj48AIHbt2lVgmw0Gg/D09BS9e/c2Of79998LAOLAgQNCCCE2b95c7CG3vGR/zrwe/v7+crl9+/YJAMLT09NkqCK7fUuWLBFCCPHw4UPh6uoqGjVqZDI0s23bNgFAzJo1Sz7WoUMHYW9vL65du5brWjzevmHDhpmU6dWrl6hevbr8fNGiRcUevgoKCso1LLVlyxYBQPzvf/8zOd6nTx8hSZK4dOmSfAyA0Gg04vz580V639TUVFGrVi0BQLRo0cJkmCchIUEAED179sy3jhdffFEAkH8m2b/7v/76q7hz5464ceOG2LRpk3BxcRE6nU7cuHEj3/rOnDkjAIiJEyfmOnfv3j1x584d+WE8fJvXcGtYWJiQJMnk5/ukYanffvtNABDr1683Ob5r1648jxMZ47AUVSrLli1DRESEyWPnzp25yvXu3TtXt3+20aNHmzzfsWMHAGDSpEkmx998800AyPWvaD8/PwQHBxfYVkmS0LdvX+zYsQPJycny8e+++w6enp7ykFh2r8e2bduQkZFRYL2F9cMPP+S6VmvWrMlVbvDgwSZDFX369IGHh4d8XU6cOIH4+Hi89tprJhNXe/TogXr16snX586dOzhw4ACGDRsGb29vk/cwHqrI9vjP4emnn8a9e/eg1+sB5FyXn376CQaDoRhXwNSOHTtgYWGB8ePHmxx/8803IYTI9XsUFBSEBg0aFOk9rKys4OjoCADo1KmTyTBPUlISAOQ5LGQs+3z2dcjWuXNnuLi4wMvLC3369EGVKlWwdevWAntjsuvJa3J5rVq14OLiIj+2bt0qn8vutQMeDefdvXsXbdu2hRACp0+fzvc9gUc9P46OjujSpQvu3r0rP1q0aAE7Ozvs27evwDqo8uKwFFUqrVq1QsuWLQssl9eKqiedu3btGjQaTa4VV+7u7nBycsK1a9cKXffjXn75ZSxevBhbt27FwIEDkZycjB07duDVV1+Vv/CDgoLQu3dvzJkzB4sWLcIzzzyDkJAQDBw4EDqdrtDv9bgOHTrA2dm5wHJ169Y1eS5JEurUqSMPe2R//ryG3+rVq4eDBw8CgLyUuFGjRoVq3+MBKHt45Z9//oGDgwNefvllfP755xgxYgTefvttdOrUCS+99BL69OkDjabo/667du0aatSokStc1K9fXz5vrCg/52xLlizB6dOn0ahRI3zyyScYOXKk/HuV/b7ZIedJnhSCli1bhv/85z9ITEzEl19+iQMHDhTq9yO7HuOAne2nn35CRkYGzpw5I88Vynb9+nXMmjULW7duzTVHJjExscD3/euvv5CYmAhXV9c8z8fHxxdYB1VeDDdEeTD+V2dhz+XVu1DUuh/Xpk0b+Pr64vvvv8fAgQPx888/48GDB3j55ZdN3nfTpk04evQofv75Z+zevRvDhg3DwoULcfTo0Tz/xV0RPGnJsfh3HpSNjQ0OHDiAffv2Yfv27di1axe+++47PPvss/jll1+KvGS5qIrycwaAGzduYPbs2QgJCcFnn32GevXqYezYsdi9ezcAwNHRER4eHjh79my+9Zw9exaenp5wcHAwOW4c7ENCQtC+fXsMHDgQMTEx+f6O1KlTB1qtFtHR0bnOBQUFAXg0F8tYVlYWunTpgvv372Pq1KmoV68eqlSpgps3b2LIkCGF6kkzGAxwdXXF+vXr8zz/pJ5VIoA38SMqMR8fHxgMBvz1118mx+Pi4pCQkAAfH58S1d+vXz/s2rULer0e3333HXx9fdGmTZtc5dq0aYN58+bhxIkTWL9+Pc6fP4/w8PASvXdhPP65hRC4dOmSPIE2+/PHxMTkem1MTIx8vlatWgCQ55docWk0GnTq1Akff/wxLly4gHnz5mHv3r3FGtLw8fHBrVu3cvWc/PHHH/L5khg3bhwA4JNPPoGHhwfmzZuHX375xeRn+Pzzz+PKlStyb9fjfvvtN1y9ehXPP/98vu9lYWGBsLAw3Lp1C0uXLs23bJUqVfDMM89g//79hb7p37lz5/Dnn39i4cKFmDp1Knr27InOnTujRo0auco+6R8FtWvXxr1799CuXTt07tw516NJkyaFagtVTgw3RCXUvXt3AMDixYtNjn/88ccAHs0tKYmXX34Z6enpWLduHXbt2oV+/fqZnP/nn39yLUVu2rQpAJgsRb98+XKhlv0W1VdffWXyhb9p0ybcvn0b3bp1AwC0bNkSrq6uWLFihUl7du7ciYsXL8rXx8XFBR06dMCXX36J69evm7zH45+vMPJa7ZXXdSms7t27IysrK1cYWLRoESRJkj9vcWzevBlbt27F3Llz4eXlBeDRUvAWLVpg0qRJ8ryXyZMnw8bGBq+++iru3btnUsf9+/cxevRo2NraYvLkyQW+5zPPPINWrVph8eLFSEtLy7fsrFmzkJWVhVdeeSXP4anHfz7ZvWLGx4UQWLJkSa7XZt+cMCEhweR4v379kJWVhffeey/XazIzM3OVJzLGYSmqVHbu3Cn/S9tY27Zt5Z6DomrSpAlCQ0OxatUqJCQkICgoCL///jvWrVuHkJAQdOzYsURtbt68OerUqYMZM2YgPT3dZEgKANatW4fPPvsMvXr1Qu3atZGUlITVq1fDwcFBDl7AowmqQM4S4IJs2rQpz+GKLl26wM3NTX5erVo1tG/fHkOHDkVcXBwWL16MOnXqYOTIkQAAS0tLfPDBBxg6dCiCgoIwYMAAeSm4r6+vydYGn3zyCdq3b4/mzZtj1KhR8t13t2/fjqioqMJeMgCP7sh74MAB9OjRAz4+PoiPj8dnn32GmjVrypOxi+KFF15Ax44dMWPGDFy9ehVNmjTBL7/8gp9++gkTJ05E7dq1i1wn8GiOzPjx49GsWTOTycoajQYrVqxA69atMWPGDHz66aeoW7cu1q1bh0GDBqFx48a57lB89+5dfPvtt4Vuy+TJk9G3b1+sXbs21wRtY08//TSWLl2K119/HXXr1pXvUPzw4UP8+eefWL9+PaysrODu7g7g0Vyq2rVr46233sLNmzfh4OCAH374Ic/707Ro0QIAMH78eAQHB8PCwgL9+/dHUFAQXn31VYSFhSEqKgrPPfccLC0t8ddff2Hjxo1YsmQJ+vTpU5RLTZWJ2dZpEZWh/JaCAxBr1qwRQuQsBV+wYEGuOvK7M25GRoaYM2eO8PPzE5aWlsLLy0tMmzZNpKWlmZTz8fERPXr0KHL7Z8yYIQCIOnXq5Dp36tQpMWDAAOHt7S10Op1wdXUVzz//vDhx4kSu9/bx8SnwvfJbCg5A7Nu3TwiRsxT822+/FdOmTROurq7CxsZG9OjRI9dSbiGE+O6770SzZs2ETqcT1apVE4MGDRL/93//l6tcdHS06NWrl3BychLW1tbC399fzJw5M1f7Hv85ZP+Mr1y5IoQQYs+ePaJnz56iRo0awsrKStSoUUMMGDBA/PnnnwVeg7yWggshRFJSknjjjTdEjRo1hKWlpahbt65YsGCByVJ1IR4tBR87dmyB7yOEEBMmTBAajUb8/vvveZ4fN26c0Gg0Jj/Ps2fPigEDBggPDw9haWkp3N3dxYABA8S5c+dyvT6/u3NnZWWJ2rVri9q1axfq7sKnT58WgwcPFt7e3sLKykpUqVJFBAQEiDfffNNkKbwQQly4cEF07txZ2NnZCWdnZzFy5Eh5WXn2f29CCJGZmSlef/114eLiIiRJyrUsfNWqVaJFixbCxsZG2Nvbi8aNG4spU6aIW7duFdheqrwkIYrR30tElV5kZCQ6duyIjRs38l/QRFSucM4NERERVSgMN0RERFShMNwQERFRhcI5N0RERFShsOeGiIiIKhSGGyIiIqpQKt1N/AwGA27dugV7e/tC7wVERERE5iWEQFJSEmrUqFHg5reVLtzcunVLvr05ERERqcuNGzdQs2bNfMtUunBjb28P4NHFeXzXXCIiIiqf9Ho9vLy85O/x/FS6cJM9FOXg4MBwQ0REpDKFmVLCCcVERERUoTDcEBERUYXCcENEREQVSqWbc1NYWVlZyMjIMHczKA9WVlYFLgMkIqLKi+HmMUIIxMbGIiEhwdxNoSfQaDTw8/ODlZWVuZtCRETlEMPNY7KDjaurK2xtbXmjv3Im+yaMt2/fhre3N38+RESUC8ONkaysLDnYVK9e3dzNoSdwcXHBrVu3kJmZCUtLS3M3h4iIyhlOXDCSPcfG1tbWzC2h/GQPR2VlZZm5JUREVB4x3OSBQx3lG38+RESUH4YbIiIiqlDKTbiZP38+JEnCxIkT8y23ceNG1KtXD9bW1mjcuDF27NhRNg0kIiIiVSgX4eb48eNYuXIlAgIC8i13+PBhDBgwAMOHD8fp06cREhKCkJAQREdHl1FLy7c7d+5gzJgx8Pb2hk6ng7u7O4KDg3Ho0CEAj4ZztmzZYt5GEhERlTKzr5ZKTk7GoEGDsHr1avzvf//Lt+ySJUvQtWtXTJ48GQDw3nvvISIiAkuXLsWKFSvKornlWu/evfHw4UOsW7cOtWrVQlxcHPbs2YN79+6Zu2mqkpFlQJw+zdzNICJSLSutBq721mZ7f7OHm7Fjx6JHjx7o3LlzgeHmyJEjmDRpksmx4ODgfHsj0tPTkZ6eLj/X6/Ulam95lZCQgN9++w2RkZEICgoCAPj4+KBVq1YAAF9fXwBAr1695HNXr14FAPz000+YM2cOLly4gBo1aiA0NBQzZsyAVvvo10OSJHz22WfYunUrIiMj4eHhgQ8//BB9+vQp2w9ZBoQQeOHTg/gjNsncTSEiUq3m3k748bV2Znt/s4ab8PBwnDp1CsePHy9U+djYWLi5uZkcc3NzQ2xs7BNfExYWhjlz5hS7jUIIPMgwz5JjG0uLQq8MsrOzg52dHbZs2YI2bdpAp9OZnD9+/DhcXV2xZs0adO3aFRYWFgCA3377DYMHD8Ynn3yCp59+GpcvX8aoUaMAALNnz5ZfP3PmTMyfPx9LlizB119/jf79++PcuXOoX7++Qp+2fEjPNMjBxkqrAddlEREVnaWFeWe9mC3c3LhxAxMmTEBERASsrUuv62ratGkmvT16vR5eXl6Ffv2DjCw0mLW7NJpWoAtzg2FrVbgfkVarxdq1azFy5EisWLECzZs3R1BQEPr374+AgAC4uLgAAJycnODu7i6/bs6cOXj77bcRGhoKAKhVqxbee+89TJkyxSTc9O3bFyNGjACQMxz46aef4rPPPlPq45Y7p2Z2gZ3O7J2bRERURGaLVidPnkR8fDyaN28OrVYLrVaL/fv345NPPoFWq83zBm3u7u6Ii4szORYXF2fyZf04nU4HBwcHk0dF1bt3b9y6dQtbt25F165dERkZiebNm2Pt2rVPfM2ZM2cwd+5cuefHzs4OI0eOxO3bt5GamiqXCwwMNHldYGAgLl68WFofxWyEyPk7e22IiNTJbP8s7dSpE86dO2dybOjQoahXrx6mTp0qD5sYCwwMxJ49e0yWi0dEROT64lWSjaUFLswNLrX6C3rvorK2tkaXLl3QpUsXzJw5EyNGjMDs2bMxZMiQPMsnJydjzpw5eOmll/Ksq7IRyEk3vFcgEZE6mS3c2Nvbo1GjRibHqlSpgurVq8vHBw8eDE9PT4SFhQEAJkyYgKCgICxcuBA9evRAeHg4Tpw4gVWrVpVaOyVJKvTQUHnUoEEDecK1paVlrh6x5s2bIyYmBnXq1Mm3nqNHj2Lw4MEmz5s1a6Z4e83NtOeG6YaISI3K9bf29evXodHkjJy1bdsWGzZswDvvvIPp06ejbt262LJlS66QVBndu3cPffv2xbBhwxAQEAB7e3ucOHECH374IXr27Ang0YqpPXv2oF27dtDpdKhatSpmzZqF559/Ht7e3ujTpw80Gg3OnDmD6Ohok9VrGzduRMuWLdG+fXusX78ev//+O7744gtzfdxSY5Rt2HNDRKRS5SrcREZG5vsceDSxtW/fvmXTIBWxs7ND69atsWjRIly+fBkZGRnw8vLCyJEjMX36dADAwoULMWnSJKxevRqenp64evUqgoODsW3bNsydOxcffPABLC0tUa9ePXnycLY5c+YgPDwcr732Gjw8PPDtt9+iQYMG5viopUoYd90QEZEqSaKS/d9cr9fD0dERiYmJuSYXp6Wl4cqVK/Dz86uU802eRJIkbN68GSEhIeZuCoDS/Tnp0zIQ8O4vAICY/3WFTlv0eU9ERKS8/L6/H1cutl8gKi8454aISP0YboiMGYcbZhsiIlUqV3NuqHyqTCOXJkvBzdgOIiIqPvbcEBkxGZZi1w0RkSox3BAZMVkKbrZWEBFRSTDcEBkxHoJjxw0RkTox3BAZMb2JH9MNEZEaMdwQGalEc6eJiCoshhsiI9mrpdhpQ0SkXgw3lYgkSfImmuYUGRkJSZKQkJDwxDJr166Fk5NTmbVJ9m/PDbMNEZF6MdxUIHfu3MGYMWPg7e0NnU4Hd3d3BAcH49ChQwCA27dvo1u3bmZu5aMNUG/fvg1HR0dzNyWX7FEpzrchIlIv3sSvAunduzcePnyIdevWoVatWoiLi8OePXtw7949AIC7u7uZW/iIlZVVuWnL4wR7boiIVI89NxVEQkICfvvtN3zwwQfo2LEjfHx80KpVK0ybNg0vvvgigNzDUocPH0bTpk1hbW2Nli1bYsuWLZAkCVFRUQByho92796NZs2awcbGBs8++yzi4+Oxc+dO1K9fHw4ODhg4cCBSU1PletPT0zF+/Hi4urrC2toa7du3x/Hjx+XzeQ1LrV27Ft7e3rC1tUWvXr3kQFbWOOeGiEj92HNTECGAjNSCy5UGS9tCf8va2dnBzs4OW7ZsQZs2baDT6fItr9fr8cILL6B79+7YsGEDrl27hokTJ+ZZ9t1338XSpUtha2uLfv36oV+/ftDpdNiwYQOSk5PRq1cvfPrpp5g6dSoAYMqUKfjhhx+wbt06+Pj44MMPP0RwcDAuXbqEatWq5ar/2LFjGD58OMLCwhASEoJdu3Zh9uzZhfrcSsvpuWG6ISJSK4abgmSkAu/XMM97T78FWFUpVFGtVou1a9di5MiRWLFiBZo3b46goCD0798fAQEBucpv2LABkiRh9erVsLa2RoMGDXDz5k2MHDkyV9n//e9/aNeuHQBg+PDhmDZtGi5fvoxatWoBAPr06YN9+/Zh6tSpSElJwfLly7F27Vp5fs/q1asRERGBL774ApMnT85V/5IlS9C1a1dMmTIFAPCf//wHhw8fxq5duwp3nRQkrwRntiEiUi0OS1UgvXv3xq1bt7B161Z07doVkZGRaN68OdauXZurbExMDAICAmBtbS0fa9WqVZ71GocjNzc32NraysEm+1h8fDwA4PLly8jIyJDDEABYWlqiVatWuHjxYp71X7x4Ea1btzY5FhgYWPAHLgXZdyhmtiEiUi/23BTE0vZRD4q53ruIrK2t0aVLF3Tp0gUzZ87EiBEjMHv2bAwZMqT4zbC0lP8uSZLJ8+xjBoOh2PWXJ/KwFNMNEZFqseemIJL0aGjIHA8FvmEbNGiAlJSUXMf9/f1x7tw5pKeny8eMJ/0WV+3atWFlZSUvPweAjIwMHD9+HA0aNMjzNfXr18exY8dMjh09erTEbSkJzrkhIlIvhpsK4t69e3j22WfxzTff4OzZs7hy5Qo2btyIDz/8ED179sxVfuDAgTAYDBg1ahQuXryI3bt346OPPgJQsnu8VKlSBWPGjMHkyZOxa9cuXLhwASNHjkRqaiqGDx+e52vGjx+PXbt24aOPPsJff/2FpUuXmmW+DcCeGyKiioDhpoKws7ND69atsWjRInTo0AGNGjXCzJkzMXLkSCxdujRXeQcHB/z888+IiopC06ZNMWPGDMyaNQsATObhFMf8+fPRu3dv/Pe//0Xz5s1x6dIl7N69G1WrVs2zfJs2bbB69WosWbIETZo0wS+//IJ33nmnRG0oLnkpuFnenYiIlCAJUbm2CtTr9XB0dERiYiIcHBxMzqWlpeHKlSvw8/Mr8Re8Gq1fvx5Dhw5FYmIibGxszN2cJyrNn9OVuyno+FEk7HRaRM8JVrRuIiIqvvy+vx/HCcWV2FdffYVatWrB09MTZ86cwdSpU9GvX79yHWxKG1dLERGpH8NNJRYbG4tZs2YhNjYWHh4e6Nu3L+bNm2fuZplVzt5SZm0GERGVAMNNJTZlyhT5xnn0SM6EYqYbIiK14oRiIhPcW4qISO0YboiMcFdwIiL1Y7ghMpIz54bxhohIrRhuiIyw54aISP0YboiMCM65ISJSPYYbIiM5t7RkuiEiUiuGGyIj3FuKiEj9GG4qiCFDhkCSJEiSBEtLS/j5+WHKlClIS0sr1OuvXr0KSZIQFRWV61xkZCQkSUJCQkKuc76+vli8eHHJGl+OcG8pIiL14038KpCuXbtizZo1yMjIwMmTJxEaGgpJkvDBBx+Yu2mqwZ4bIiL1Y89NBaLT6eDu7g4vLy+EhISgc+fOiIiIAAAYDAaEhYXBz88PNjY2aNKkCTZt2mTmFpdfEvtuiIhUiz03BRBC4EHmA7O8t43Wptj3W4mOjsbhw4fh4+MDAAgLC8M333yDFStWoG7dujhw4ABeeeUVuLi4ICgoSMlmqxp7boiI1I/hpgAPMh+g9YbWZnnvYwOPwdbSttDlt23bBjs7O2RmZiI9PR0ajQZLly5Feno63n//ffz6668IDAwEANSqVQsHDx7EypUrGW6McM4NEZH6MdxUIB07dsTy5cuRkpKCRYsWQavVonfv3jh//jxSU1PRpUsXk/IPHz5Es2bNzNTa8okbZxIRqR/DTQFstDY4NvCY2d67KKpUqYI6deoAAL788ks0adIEX3zxBRo1agQA2L59Ozw9PU1eo9PpCqzXwcEBAJCYmAgnJyeTcwkJCXB0dCxSO8szUXARIiIq5xhuCiBJUpGGhsoLjUaD6dOnY9KkSfjzzz+h0+lw/fr1Yg1B1a1bFxqNBidPnpTn8ADA33//jcTERPznP/9RsulmJQTvUExEpHYMNxVY3759MXnyZKxcuRJvvfUW3njjDRgMBrRv3x6JiYk4dOgQHBwcEBoaKr8mJiYmVz0NGzbEiBEj8Oabb0Kr1aJx48a4ceMGpk6dijZt2qBt27Zl+bFKVc7GmWZtBhERlQDDTQWm1Woxbtw4fPjhh7hy5QpcXFwQFhaGv//+G05OTmjevDmmT59u8pr+/fvnqufGjRtYsmQJ5s+fj6lTp+LatWtwd3dHly5dMG/evAo1PyVn48yK85mIiCobSQhRqaYZ6PV6ODo6IjExUZ5Lki0tLQ1XrlyBn58frK2tzdRCKkhp/pxOXruP3suPwKe6LfZP7qho3UREVHz5fX8/jjfxIzKS03NDRERqxXBDZCRnzg3jDRGRWpk13CxfvhwBAQFwcHCAg4MDAgMDsXPnzieWX7t2rbw5ZPaDw0ekJPbcEBGpn1knFNesWRPz589H3bp1IYTAunXr0LNnT5w+fRoNGzbM8zUODg4mK3r4L2xSkmC6ISJSPbOGmxdeeMHk+bx587B8+XIcPXr0ieFGkiS4u7uXarsq2Rxr1SnNn488LFVq70BERKWt3My5ycrKQnh4OFJSUuT9j/KSnJwMHx8feHl5oWfPnjh//ny+9aanp0Ov15s8nsTS0hIAkJqaWrwPQWXi4cOHAAALCwvF6+b2C0RE6mf2+9ycO3cOgYGBSEtLg52dHTZv3owGDRrkWdbf3x9ffvklAgICkJiYiI8++ght27bF+fPnUbNmzTxfExYWhjlz5hSqLRYWFnByckJ8fDwAwNbWll9y5YzBYMCdO3dga2sLrVb5X19unElEpH5mv8/Nw4cPcf36dSQmJmLTpk34/PPPsX///icGHGMZGRmoX78+BgwYgPfeey/PMunp6UhPT5ef6/V6eHl5PXGdvBACsbGxSEhIKPZnotKl0Wjg5+cHKysrxes+fOkuBn5+DP9xs8Mvb3C3dCKi8qIo97kxe8+NlZWVvNljixYtcPz4cSxZsgQrV64s8LWWlpZo1qwZLl269MQyOp2uUJtDZpMkCR4eHnB1dUVGRkahX0dlx8rKChpN6Yyo5sy5Yd8NEZFamT3cPM5gMJj0tOQnKysL586dQ/fu3RVvh4WFRanM6aDyzcCNM4mIVM+s4WbatGno1q0bvL29kZSUhA0bNiAyMhK7d+8GAAwePBienp4ICwsDAMydOxdt2rRBnTp1kJCQgAULFuDatWsYMWKEOT8GVSCcUExEpH5mDTfx8fEYPHgwbt++DUdHRwQEBGD37t3o0qULAOD69esmww///PMPRo4cidjYWFStWhUtWrTA4cOHCzU/h6gwuBSciEj9zD6huKwVZUISVT6RMfEYsuY4GtZwwPbxT5u7OURE9C9unElUTDl7S5m1GUREVAIMN0TG5N0XmG6IiNSK4YbIiHwTP2YbIiLVYrghMsJ9M4mI1I/hhsiI4KQbIiLVY7ghMsKl4ERE6sdwQ2RE8A7FRESqx3BDZIQ9N0RE6sdwQ2SE2y8QEakfww2RiX+HpczcCiIiKj6GGyIjOT035m0HEREVH8MNkZGcOTdMN0REasVwQ2REcEYxEZHqMdwQGRGcc0NEpHoMN0RGOOeGiEj9GG6IjHDODRGR+jHcEBnhHYqJiNSP4YYoDww3RETqxXBDZESec8NhKSIi1WK4ITIir5ZitiEiUi2GGyIj8n1uiIhItRhuiIxw40wiIvVjuCEywhsUExGpH8MNkREuBSciUj+GGyIj7LkhIlI/hhsiY5xzQ0Skegw3REa4cSYRkfox3BAZMbDnhohI9RhuiIxwV3AiIvVjuCEywmEpIiL1Y7ghMsKeGyIi9WO4ITKSsxSc6YaISK0YboiM8SZ+RESqx3BDZETuuWG4ISJSLYYbIiPynBsOSxERqRbDDZERkZNuiIhIpRhuiIxwbykiIvVjuCEyIniHYiIi1WO4ITLCnhsiIvVjuCEyIrgUnIhI9RhuiPLAbENEpF4MN0RGOOeGiEj9GG6IjHDjTCIi9WO4ITIiOKOYiEj1GG6IjHDjTCIi9TNruFm+fDkCAgLg4OAABwcHBAYGYufOnfm+ZuPGjahXrx6sra3RuHFj7Nixo4xaS5VBzpwb87aDiIiKz6zhpmbNmpg/fz5OnjyJEydO4Nlnn0XPnj1x/vz5PMsfPnwYAwYMwPDhw3H69GmEhIQgJCQE0dHRZdxyqqg454aISP0kIW+mUz5Uq1YNCxYswPDhw3Ode/nll5GSkoJt27bJx9q0aYOmTZtixYoVhapfr9fD0dERiYmJcHBwUKzdVDEs23cJC3bHoF/LmviwTxNzN4eIiP5VlO/vcjPnJisrC+Hh4UhJSUFgYGCeZY4cOYLOnTubHAsODsaRI0eeWG96ejr0er3Jg6ggnHNDRKReZg83586dg52dHXQ6HUaPHo3NmzejQYMGeZaNjY2Fm5ubyTE3NzfExsY+sf6wsDA4OjrKDy8vL0XbTxUL71BMRKR+Zg83/v7+iIqKwrFjxzBmzBiEhobiwoULitU/bdo0JCYmyo8bN24oVjdVPJxQTESkflpzN8DKygp16tQBALRo0QLHjx/HkiVLsHLlylxl3d3dERcXZ3IsLi4O7u7uT6xfp9NBp9Mp22iqsHImoDHdEBGpldl7bh5nMBiQnp6e57nAwEDs2bPH5FhERMQT5+gQFRV7boiI1M+sPTfTpk1Dt27d4O3tjaSkJGzYsAGRkZHYvXs3AGDw4MHw9PREWFgYAGDChAkICgrCwoUL0aNHD4SHh+PEiRNYtWqVOT8GVSBcCk5EpH5mDTfx8fEYPHgwbt++DUdHRwQEBGD37t3o0qULAOD69evQaHI6l9q2bYsNGzbgnXfewfTp01G3bl1s2bIFjRo1MtdHoAqGPTdEROpn1nDzxRdf5Hs+MjIy17G+ffuib9++pdQiquy4/QIRkfqVuzk3RGbFpeBERKrHcENkxJA9LGXeZhARUQkw3BAZkScUs+uGiEi1GG6IjHBCMRGR+jHcEBnhhGIiIvVjuCEywp4bIiL1Y7ghMsKb+BERqR/DDZEx9twQEakeww2REXnODdMNEZFqMdwQGRGCw1JERGrHcENkROQslyIiIpViuCEywqXgRETqx3BDZIRLwYmI1I/hhsgIl4ITEakfww2REfbcEBGpH8MNUR4454aISL0YboiMyEvBmW2IiFSL4YbICFeCExGpH8MNkZGc+9ww3hARqRXDDZERrpYiIlI/hhsiI1wtRUSkfgw3REZ4h2IiIvVjuCEywp4bIiL1Y7ghMsE5N0REasdwQ2SEPTdEROrHcENkJCfcMN0QEakVww2RESFPKSYiIrViuCEywmEpIiL1Y7ghMsKl4ERE6sdwQ2SEPTdEROrHcENkhNsvEBGpH8MNkTH23BARqR7DDZERw7/jUhqmGyIi1WK4ITLCheBEROrHcENkhDfxIyJSvyKHm4yMDGi1WkRHR5dGe4jMKmcpOBERqVWRw42lpSW8vb2RlZVVGu0hMivxb9cNO26IiNSrWMNSM2bMwPTp03H//n2l20NkVuy5ISJSP21xXrR06VJcunQJNWrUgI+PD6pUqWJy/tSpU4o0jqjMcc4NEZHqFSvchISEKNwMovJBvokfsw0RkWoVK9zMnj1b6XYQlQvyainzNoOIiEqgWOEm28mTJ3Hx4kUAQMOGDdGsWTNFGkVkLkKedMN4Q0SkVsUKN/Hx8ejfvz8iIyPh5OQEAEhISEDHjh0RHh4OFxcXJdtIVGa4txQRkfoVa7XU66+/jqSkJJw/fx7379/H/fv3ER0dDb1ej/HjxyvdRqIyw13BiYjUr1g9N7t27cKvv/6K+vXry8caNGiAZcuW4bnnnlOscURlLWcpONMNEZFaFavnxmAwwNLSMtdxS0tLGAyGQtcTFhaGp556Cvb29nB1dUVISAhiYmLyfc3atWshSZLJw9rausifgSgv7LkhIlK/YoWbZ599FhMmTMCtW7fkYzdv3sQbb7yBTp06Fbqe/fv3Y+zYsTh69CgiIiKQkZGB5557DikpKfm+zsHBAbdv35Yf165dK87HIMoD59wQEaldsW/i9+KLL8LX1xdeXl4AgBs3bqBRo0b45ptvCl3Prl27TJ6vXbsWrq6uOHnyJDp06PDE10mSBHd39+I0nShf7LkhIlK/YoUbLy8vnDp1Cr/++iv++OMPAED9+vXRuXPnEjUmMTERAFCtWrV8yyUnJ8PHxwcGgwHNmzfH+++/j4YNG+ZZNj09Henp6fJzvV5fojZSxcY5N0RE6lfkcJORkQEbGxtERUWhS5cu6NKliyINMRgMmDhxItq1a4dGjRo9sZy/vz++/PJLBAQEIDExER999BHatm2L8+fPo2bNmrnKh4WFYc6cOYq0kSo+wbv4ERGpXrnZFXzs2LGIjo5GeHh4vuUCAwMxePBgNG3aFEFBQfjxxx/h4uKClStX5ll+2rRpSExMlB83btxQtN1UsXDjTCIi9SsXu4KPGzcO27Ztw759+/LsfcmPpaUlmjVrhkuXLuV5XqfTwcHBweRB9CSCG2cSEameWXcFF0Lg9ddfx+bNmxEZGQk/P78ityUrKwvnzp1D9+7di/xaosex54aISP3Muiv42LFjsWHDBvz000+wt7dHbGwsAMDR0RE2NjYAgMGDB8PT0xNhYWEAgLlz56JNmzaoU6cOEhISsGDBAly7dg0jRoxQpE1UuWXPuWHHDRGRehU53GRmZkKSJAwbNqzIQ0iPW758OQDgmWeeMTm+Zs0aDBkyBABw/fp1aDQ5o2f//PMPRo4cidjYWFStWhUtWrTA4cOH0aBBgxK1hcgYww0RkXpJQl4eUnj29vY4d+4cfH19S6FJpUuv18PR0RGJiYmcf0O5vPL5MRy8dBeLX26KkGae5m4OERH9qyjf38W+Q/H+/fuL1Tii8kzeFZw9N0REqlWsOTfdunXD22+/jXPnzqFFixa5JhS/+OKLijSOqKwVvR+TiIjKm2KFm9deew0A8PHHH+c6J0mS4vfAISorXApORKR+xQo3Rdn5m0hNBDfOJCJSvSLNuenevbu8/xMAzJ8/HwkJCfLze/fucdUSqRo3ziQiUr8ihZvdu3ebbEL5/vvvm9ylODMzEzExMcq1jqiMceNMIiL1K1K4eXzVeDFWkROVa9m/0xpmGyIi1SrWUnCiiorDUkRE6lekcCNJUq5VJFxVQhVJTl8kf6+JiNSqSKulhBAYMmQIdDodACAtLQ2jR4+W73NjPB+HSI24txQRkfoVKdyEhoaaPH/llVdylRk8eHDJWkRkRtwVnIhI/YoUbtasWVNa7SAqF3gTPyIi9eOEYiIj7LkhIlI/hhsiY5xzQ0Skegw3REbknhuGGyIi1WK4ITIiz7nhwBQRkWox3BAZyd44k9mGiEi9GG6IjAhmGyIi1WO4ITLCpeBEROrHcENkhEvBiYjUj+GGyAi3XyAiUj+GG6I8cLUUEZF6MdwQGcmZc2PedhARUfEx3BAZyV4KzmxDRKReDDdERgRnFBMRqR7DDZGRnGzDdENEpFYMN0RGuFqKiEj9GG6IjHBUiohI/RhuiIzxDsVERKrHcENkRO65YbYhIlIthhsiI/KcGzO3g4iIio/hhsgIe26IiNSP4YbIiHyfG/bdEBGpFsMNkRH5DsXMNkREqsVwQ2RE3lvKvM0gIqISYLghMiK4FJyISPUYbojywGhDRKReDDdERgz/dt1o2HNDRKRaDDdERnKGpczbDiIiKj6GGyIjQr7TDRERqRXDDZER9twQEakfww2RkZxdwZluiIjUiuGGyAh7boiI1I/hhsgE71BMRKR2DDdERnLuUMx0Q0SkVmYNN2FhYXjqqadgb28PV1dXhISEICYmpsDXbdy4EfXq1YO1tTUaN26MHTt2lEFrqTLgruBEROpn1nCzf/9+jB07FkePHkVERAQyMjLw3HPPISUl5YmvOXz4MAYMGIDhw4fj9OnTCAkJQUhICKKjo8uw5VRRiX+7bphtiIjUSxLZ/zcvB+7cuQNXV1fs378fHTp0yLPMyy+/jJSUFGzbtk0+1qZNGzRt2hQrVqwo8D30ej0cHR2RmJgIBwcHxdqenpmFO0npitWXTae1gIu9TrH6HjzMAvDofi7GP3lJynsoxrhcdpn0zCxoLTTIyhJIz8qCTmsBC42EjEwDMg0C1pYaSJKE9IwsCADWlhYQQiA90wAJgM7SAgYhkJ5hgFYjwVKrQZZBID0zCzoLC1hYSMjMMiAjS0BnqYFGkvAw04Aso7rTMh59juy60zIMsNBIsNJq5LotLaR822lct0EI6LQatPjfr3iYacCvkzqgjqu9YtediIhKpijf39oyalOhJCYmAgCqVav2xDJHjhzBpEmTTI4FBwdjy5YteZZPT09HenpO6NDr9SVvaB7O39Ljpc8Ol0rdc15siNC2viWu54uDV/Detgslb1ClwL4bIiK1KjfhxmAwYOLEiWjXrh0aNWr0xHKxsbFwc3MzOebm5obY2Ng8y4eFhWHOnDmKtjUvEgCdVtlRvkyDQJZB4MyNBEXqY7ApPM65ISJSr3ITbsaOHYvo6GgcPHhQ0XqnTZtm0tOj1+vh5eWl6HsAgJdLFuaHpsHG0gZdfbsqUueqA5fx/o4/uCGAGTDbEBGpV7kIN+PGjcO2bdtw4MAB1KxZM9+y7u7uiIuLMzkWFxcHd3f3PMvrdDrodMrNWXmS/0v6P8w6PAs17WoqFm6y58CUo2lRlYbErhsiItUy62opIQTGjRuHzZs3Y+/evfDz8yvwNYGBgdizZ4/JsYiICAQGBpZWMwtFIz26lAZhUKzO7O9XRpuyx2hDRKReZu25GTt2LDZs2ICffvoJ9vb28rwZR0dH2NjYAAAGDx4MT09PhIWFAQAmTJiAoKAgLFy4ED169EB4eDhOnDiBVatWme1zAICFZAEAyBJZitfNjpuyx44bIiL1MmvPzfLly5GYmIhnnnkGHh4e8uO7776Ty1y/fh23b9+Wn7dt2xYbNmzAqlWr0KRJE2zatAlbtmzJdxJyWcjuuVFyCCl7aITZpuzxDsVEROpl1p6bwgSByMjIXMf69u2Lvn37lkKLii873CjZc5P99co5N2WPPTdEROrFvaUUwjk3RERE5QPDjUKy59wYoGC4yf4L002ZY88NEZF6MdwoJHt+jMGgZM9N9pwbppuyxqXgRETqxXCjkNJYLSUPSzHblDlGGyIi9WK4UYi8WkrBXpacCcWKVUmFxI4bIiL1YrhRiNxzY1DwPjccljIbLgUnIlIvhhuFyHNulFwt9e+f7Lkpe+y5ISJSL4YbhZTKaikuBTcbZhsiIvViuFGI8X1ulLrpXs7GmYpUR0XBdENEpFoMNwrJDjeAckNTEm90Yzacc0NEpF4MNwoxCTcKDU1xzo35cM4NEZF6MdwoJHvODaB8zw2zTdnTMN0QEakWw41CjHtulFoOnjPnhvGmrDHaEBGpF8ONQkpjzg3Yc2M27LghIlIvhhuFmAxLcc6N6nFCMRGRejHcKMSk50ahzTNzNs6kMsdsQ0SkWgw3CjGZc6PQ5pk5PTeMN2WNw1JEROrFcKMQSZJyJgAr1NfCL1jz4aUnIlIvhhsFKb15prwUnB03ZU5isiQiUi2GGwUpvXmm0j1BVHiMNkRE6sVwoyClN89kz435sOOGiEi9GG4UJG+eqfRqKYabMsel4ERE6sVwo6DscKP4aikOS5U59twQEakXw42C5J4bDksRERGZDcONguQ5N0oNS4E38TMX9twQEakXw42CFB+WyhmXojLGOTdEROrFcKOg7HCj2E38/v2Tc27KHntuiIjUi+FGQaXVc2NgtilzzDZEROrFcKMgpefcZH/Fcm+pssc7FBMRqRfDjYLkOxQrvVpKkdqoKBhtiIjUi+FGQXLPjWLbLzzCjpuyx44bIiL1YrhRkDznRrGNM7kU3Fw4LEVEpF4MNwrK7rlRerUUu26IiIgKj+FGQdn/2ld6tRSjDRERUeEx3ChI8TsUc/sFIiKiImO4UZDie0vJ2y8w3RARERUWw42C5HCj0GopsOeGiIioyBhuFJQ9LKXYaql//2S4ISIiKjyGGwUp3XPDpeBERERFx3CjIAvNo56bTJGpSH05PTeMN0RERIXFcKMgrUYLAMgwZChSH+8jR0REVHQMNwrKDjeZBqV6brI3zlSkOiIiokqB4UZBlhpLAAqGG/kmfkw3REREhcVwoyDFw82/f7LnhoiIqPAYbhSklZQdlgK3XyAiIioys4abAwcO4IUXXkCNGjUgSRK2bNmSb/nIyEhIkpTrERsbWzYNLkDpzblhvCEiIioss4ablJQUNGnSBMuWLSvS62JiYnD79m354erqWkotLBo53Ci1FJw9N0REREWmNeebd+vWDd26dSvy61xdXeHk5KR8g0pI6aXgGqYbIiKiIlPlnJumTZvCw8MDXbp0waFDh/Itm56eDr1eb/IoLYoPSzHbEBERFZmqwo2HhwdWrFiBH374AT/88AO8vLzwzDPP4NSpU098TVhYGBwdHeWHl5dXqbVP+Tk3j3DODRERUeGZdViqqPz9/eHv7y8/b9u2LS5fvoxFixbh66+/zvM106ZNw6RJk+Tner2+1AIOe26IiIjMT1XhJi+tWrXCwYMHn3hep9NBp9OVSVsUXwrOOxQTEREVmaqGpfISFRUFDw8PczcDAO9QTEREVB6YtecmOTkZly5dkp9fuXIFUVFRqFatGry9vTFt2jTcvHkTX331FQBg8eLF8PPzQ8OGDZGWlobPP/8ce/fuxS+//GKuj2BC8aXg//7JnhsiIqLCM2u4OXHiBDp27Cg/z54bExoairVr1+L27du4fv26fP7hw4d48803cfPmTdja2iIgIAC//vqrSR3mpPycGw5LERERFZVZw80zzzyT70qgtWvXmjyfMmUKpkyZUsqtKj75PjdZytznRiq4CBERET1G9XNuypNSu0Mxu26IiIgKjeFGQUrfoVjeW0qR2oiIiCoHhhsFldpqKaYbIiKiQmO4UZDSE4qzcSk4ERFR4THcKKjU7lDMbENERFRoDDcKyh6W4pwbIiIi82G4UZC1hTUAID0rXZH62HNDRERUdAw3CrLWPgo3aZlpitTHpeBERERFx3CjoOyem7QshcINh6WIiIiKjOFGQey5ISIiMj+GGwXpLHQAFAw3//7JaENERFR4DDcKkntustIU6W3hhGIiIqKiY7hRkI3WRv67MiumsncFZ7ohIiIqLIYbBWUPSwHKhBu556bENREREVUeDDcK0mq08l2KH2Q+KHF92XNumG6IiIgKj+FGYfJycAUmFUsSl4ITEREVFcONwrInFSsyLPXvn5xzQ0REVHgMNwrLnlScmpla4ro454aIiKjoGG4UZm9lDwBIephU4rrkOxQz3RARERUaw43CHK0cAQCJ6Yklriun54bphoiIqLAYbhTmoHMAAOgf6ktcF2/iR0REVHQMNwpTtueGq6WIiIiKiuFGYYr23GT/hemGiIio0BhuFMY5N0RERObFcKOw6jbVAQB3Uu+UuC6uliIiIio6hhuFeVTxAADcSrlV4rp4nxsiIqKiY7hRmKedJwDgdsptGIShRHXxDsVERERFx3CjMBdbF2glLTINmYhLiStZZey5ISIiKjKGG4VpNVrUcqoFALhw70KJ6uKcGyIioqJjuCkFjZ0bAwDO3DlTonokqeAyREREZIrhphS09mgNAPj1+q8lmndjnG0474aIiKhwGG5KQVDNINhZ2uFG0g1s/mtzseuRjLpumG2IiIgKh+GmFNha2mJ44+EAgHnH5uGnSz8Vq+fFpOdGobYRERFVdAw3pWRYo2F4zuc5ZBgy8M6hd/Dytpfx418/FunOxcZzbjgsRUREVDhaczegotJIGnzY4UPUi66H1edW4+L9i5h9eDbeO/IeGjo3REu3lqhXrR78HP3g6+gLnYUuVx2SUd8Now0REVHhMNyUIguNBUYGjESf//TBpj83YceVHbiUcAln7pwxWUmlkTRwtnZGdZvqcLZ59KedpR000MGq+k0Igw4/XXoIa0tLaCUttBotLCQLWGgsoNVooZW0sNBYmIQh4/k62cc1NtdyGieetBSLS7QA4Nydc8V+LfcCKxlePyL1s9Xaom7VumZ7f0lUsvEOvV4PR0dHJCYmwsHBoczf/0bSDZyMO4mo+ChcSriEvxP/RtLDpDJvBxERUWlp4tIE33T/RtE6i/L9zZ6bMuZl7wUvey+E1AkB8Gguzb20e4hLjcO9B/dw98Fd3HtwD6mZqUhIS8a3x/8CNA8R5O8AA7KQJbKQZchCpshEpiETWYZHxzINmfJ7GP/LNzu7Cghcu5cql8ibMPqzcvfgeFWzNXcTKjWpkv/+Eamdq62rWd+f4cbMJEmCs40znG2cc51LTs/Emq27AQBLRneFtaVFsd/HYBCoNX1HsV9f2ewa2cPcTSAiomLiaqlyzPQmfiWrq1KNPRIRUaXGcFOOmSwFL2E8qWRTq4iIqBJjuCnHTJaCs+eGiIioUBhuyjHTnpuSYccNERFVFgw3KlHSYSXeO4SIiCoLrpZSiiELyEzL40QeS1qlvJa55j4mZRlghYxH1WekA1pDsesTWVnQIPcO5XlFHpHnMlwuzSUiInUwa7g5cOAAFixYgJMnT+L27dvYvHkzQkJC8n1NZGQkJk2ahPPnz8PLywvvvPMOhgwZUibtzdfNU8AXnRWtUgfgT+t/nywsWV3WAP62LrBYiRnyuPPx4wEqr/CUd6Ayo7nlrD1UgfB3iyoBr1bAUPPdfsSs4SYlJQVNmjTBsGHD8NJLLxVY/sqVK+jRowdGjx6N9evXY8+ePRgxYgQ8PDwQHBxcBi2mgmikwgx/qWCILHcnFxERFVZWhlnfvtxsvyBJUoE9N1OnTsX27dsRHR0tH+vfvz8SEhKwa9euQr1PqW2/kNewVJ6XNo9jTyiXZRBoOjcCAPD96DZwsM4jixbyPdIzsvDSZ4dNjkl5lMt7QKr45R5vS+FfZ17HpncydxOIiNRLYwnYuShaZYXdfuHIkSPo3Nl06Cc4OBgTJ0584mvS09ORnp4uP9fr9aXTOI0FYFVF0Solg0ASHm0D0G3FWQVqtFegjkrCoYa5W0BERMWkqtVSsbGxcHNzMznm5uYGvV6PBw8e5PmasLAwODo6yg8vL6+yaKoiNBoJzwd4QKfVKPKw0Cg31l/V1tLkuaON6XNbq5ytIiwtJJP3tteZZmqnx+p6vO7He6ysLXN+bXVajcl8avvHyj5e1+PvZdwWjQRYWWjwQe/GICIi9VJVz01xTJs2DZMmTZKf6/V6VQWcpQObm7sJREREqqKqcOPu7o64uDiTY3FxcXBwcICNjU2er9HpdNDpdGXRPCIiIioHVDUsFRgYiD179pgci4iIQGBgoJlaREREROWNWcNNcnIyoqKiEBUVBeDRUu+oqChcv34dwKMhpcGDB8vlR48ejb///htTpkzBH3/8gc8++wzff/893njjDXM0n4iIiMohs4abEydOoFmzZmjWrBkAYNKkSWjWrBlmzZoFALh9+7YcdADAz88P27dvR0REBJo0aYKFCxfi888/5z1uiIiISFZu7nNTVkrtPjdERERUaory/a2qOTdEREREBWG4ISIiogqF4YaIiIgqFIYbIiIiqlAYboiIiKhCYbghIiKiCoXhhoiIiCoUhhsiIiKqUBhuiIiIqEJR1a7gSsi+IbNerzdzS4iIiKiwsr+3C7OxQqULN0lJSQAALy8vM7eEiIiIiiopKQmOjo75lql0e0sZDAbcunUL9vb2kCRJ0br1ej28vLxw48YN7ltVinidywavc9nhtS4bvM5lo7SusxACSUlJqFGjBjSa/GfVVLqeG41Gg5o1a5bqezg4OPA/nDLA61w2eJ3LDq912eB1LhulcZ0L6rHJxgnFREREVKEw3BAREVGFwnCjIJ1Oh9mzZ0On05m7KRUar3PZ4HUuO7zWZYPXuWyUh+tc6SYUExERUcXGnhsiIiKqUBhuiIiIqEJhuCEiIqIKheGGiIiIKhSGG4UsW7YMvr6+sLa2RuvWrfH777+bu0mqEhYWhqeeegr29vZwdXVFSEgIYmJiTMqkpaVh7NixqF69Ouzs7NC7d2/ExcWZlLl+/Tp69OgBW1tbuLq6YvLkycjMzCzLj6Iq8+fPhyRJmDhxonyM11kZN2/exCuvvILq1avDxsYGjRs3xokTJ+TzQgjMmjULHh4esLGxQefOnfHXX3+Z1HH//n0MGjQIDg4OcHJywvDhw5GcnFzWH6Vcy8rKwsyZM+Hn5wcbGxvUrl0b7733nsn+Q7zWRXfgwAG88MILqFGjBiRJwpYtW0zOK3VNz549i6effhrW1tbw8vLChx9+qMwHEFRi4eHhwsrKSnz55Zfi/PnzYuTIkcLJyUnExcWZu2mqERwcLNasWSOio6NFVFSU6N69u/D29hbJyclymdGjRwsvLy+xZ88eceLECdGmTRvRtm1b+XxmZqZo1KiR6Ny5szh9+rTYsWOHcHZ2FtOmTTPHRyr3fv/9d+Hr6ysCAgLEhAkT5OO8ziV3//594ePjI4YMGSKOHTsm/v77b7F7925x6dIlucz8+fOFo6Oj2LJlizhz5ox48cUXhZ+fn3jw4IFcpmvXrqJJkybi6NGj4rfffhN16tQRAwYMMMdHKrfmzZsnqlevLrZt2yauXLkiNm7cKOzs7MSSJUvkMrzWRbdjxw4xY8YM8eOPPwoAYvPmzSbnlbimiYmJws3NTQwaNEhER0eLb7/9VtjY2IiVK1eWuP0MNwpo1aqVGDt2rPw8KytL1KhRQ4SFhZmxVeoWHx8vAIj9+/cLIYRISEgQlpaWYuPGjXKZixcvCgDiyJEjQohH/zFqNBoRGxsrl1m+fLlwcHAQ6enpZfsByrmkpCRRt25dERERIYKCguRww+usjKlTp4r27ds/8bzBYBDu7u5iwYIF8rGEhASh0+nEt99+K4QQ4sKFCwKAOH78uFxm586dQpIkcfPmzdJrvMr06NFDDBs2zOTYSy+9JAYNGiSE4LVWwuPhRqlr+tlnn4mqVaua/H9j6tSpwt/fv8Rt5rBUCT18+BAnT55E586d5WMajQadO3fGkSNHzNgydUtMTAQAVKtWDQBw8uRJZGRkmFznevXqwdvbW77OR44cQePGjeHm5iaXCQ4Ohl6vx/nz58uw9eXf2LFj0aNHD5PrCfA6K2Xr1q1o2bIl+vbtC1dXVzRr1gyrV6+Wz1+5cgWxsbEm19nR0RGtW7c2uc5OTk5o2bKlXKZz587QaDQ4duxY2X2Ycq5t27bYs2cP/vzzTwDAmTNncPDgQXTr1g0Ar3VpUOqaHjlyBB06dICVlZVcJjg4GDExMfjnn39K1MZKt3Gm0u7evYusrCyT/9EDgJubG/744w8ztUrdDAYDJk6ciHbt2qFRo0YAgNjYWFhZWcHJycmkrJubG2JjY+Uyef0css/RI+Hh4Th16hSOHz+e6xyvszL+/vtvLF++HJMmTcL06dNx/PhxjB8/HlZWVggNDZWvU17X0fg6u7q6mpzXarWoVq0ar7ORt99+G3q9HvXq1YOFhQWysrIwb948DBo0CAB4rUuBUtc0NjYWfn5+uerIPle1atVit5HhhsqdsWPHIjo6GgcPHjR3UyqcGzduYMKECYiIiIC1tbW5m1NhGQwGtGzZEu+//z4AoFmzZoiOjsaKFSsQGhpq5tZVLN9//z3Wr1+PDRs2oGHDhoiKisLEiRNRo0YNXutKjMNSJeTs7AwLC4tcq0ni4uLg7u5uplap17hx47Bt2zbs27cPNWvWlI+7u7vj4cOHSEhIMClvfJ3d3d3z/Dlkn6NHw07x8fFo3rw5tFottFot9u/fj08++QRarRZubm68zgrw8PBAgwYNTI7Vr18f169fB5BznfL7/4a7uzvi4+NNzmdmZuL+/fu8zkYmT56Mt99+G/3790fjxo3x3//+F2+88QbCwsIA8FqXBqWuaWn+v4ThpoSsrKzQokUL7NmzRz5mMBiwZ88eBAYGmrFl6iKEwLhx47B582bs3bs3V1dlixYtYGlpaXKdY2JicP36dfk6BwYG4ty5cyb/QUVERMDBwSHXF01l1alTJ5w7dw5RUVHyo2XLlhg0aJD8d17nkmvXrl2uWxn8+eef8PHxAQD4+fnB3d3d5Drr9XocO3bM5DonJCTg5MmTcpm9e/fCYDCgdevWZfAp1CE1NRUajelXmYWFBQwGAwBe69Kg1DUNDAzEgQMHkJGRIZeJiIiAv79/iYakAHApuBLCw8OFTqcTa9euFRcuXBCjRo0STk5OJqtJKH9jxowRjo6OIjIyUty+fVt+pKamymVGjx4tvL29xd69e8WJEydEYGCgCAwMlM9nL1F+7rnnRFRUlNi1a5dwcXHhEuUCGK+WEoLXWQm///670Gq1Yt68eeKvv/4S69evF7a2tuKbb76Ry8yfP184OTmJn376SZw9e1b07Nkzz6W0zZo1E8eOHRMHDx4UdevWrdTLk/MSGhoqPD095aXgP/74o3B2dhZTpkyRy/BaF11SUpI4ffq0OH36tAAgPv74Y3H69Glx7do1IYQy1zQhIUG4ubmJ//73vyI6OlqEh4cLW1tbLgUvTz799FPh7e0trKysRKtWrcTRo0fN3SRVAZDnY82aNXKZBw8eiNdee01UrVpV2Nrail69eonbt2+b1HP16lXRrVs3YWNjI5ydncWbb74pMjIyyvjTqMvj4YbXWRk///yzaNSokdDpdKJevXpi1apVJucNBoOYOXOmcHNzEzqdTnTq1EnExMSYlLl3754YMGCAsLOzEw4ODmLo0KEiKSmpLD9GuafX68WECROEt7e3sLa2FrVq1RIzZswwWV7Ma110+/bty/P/yaGhoUII5a7pmTNnRPv27YVOpxOenp5i/vz5irRfEsLoNo5EREREKsc5N0RERFShMNwQERFRhcJwQ0RERBUKww0RERFVKAw3REREVKEw3BAREVGFwnBDREREFQrDDRFVSpIkYcuWLeZuBhGVAoYbIipzQ4YMgSRJuR5du3Y1d9OIqALQmrsBRFQ5de3aFWvWrDE5ptPpzNQaIqpI2HNDRGah0+ng7u5u8sjeCViSJCxfvhzdunWDjY0NatWqhU2bNpm8/ty5c3j22WdhY2OD6tWrY9SoUUhOTjYp8+WXX6Jhw4bQ6XTw8PDAuHHjTM7fvXsXvXr1gq2tLerWrYutW7fK5/755x8MGjQILi4usLGxQd26dXOFMSIqnxhuiKhcmjlzJnr37o0zZ85g0KBB6N+/Py5evAgASElJQXBwMKpWrYrjx49j48aN+PXXX03Cy/LlyzF27FiMGjUK586dw9atW1GnTh2T95gzZw769euHs2fPonv37hg0aBDu378vv/+FCxewc+dOXLx4EcuXL4ezs3PZXQAiKj5Ftt8kIiqC0NBQYWFhIapUqWLymDdvnhDi0S7xo0ePNnlN69atxZgxY4QQQqxatUpUrVpVJCcny+e3b98uNBqNiI2NFUIIUaNGDTFjxowntgGAeOedd+TnycnJAoDYuXOnEEKIF154QQwdOlSZD0xEZYpzbojILDp27Ijly5ebHKtWrZr898DAQJNzgYGBiIqKAgBcvHgRTZo0QZUqVeTz7dq1g8FgQExMDCRJwq1bt9CpU6d82xAQECD/vUqVKnBwcEB8fDwAYMyYMejduzdOnTqF5557DiEhIWjbtm2xPisRlS2GGyIyiypVquQaJlKKjY1NocpZWlqaPJckCQaDAQDQrVs3XLt2DTt27EBERAQ6deqEsWPH4qOPPlK8vUSkLM65IaJy6ejRo7me169fHwBQv359nDlzBikpKfL5Q4cOQaPRwN/fH/b29vD19cWePXtK1AYXFxeEhobim2++weLFi7Fq1aoS1UdEZYM9N0RkFunp6YiNjTU5ptVq5Um7GzduRMuWLdG+fXusX78ev//+O7744gsAwKBBgzB79myEhobi3XffxZ07d/D666/jv//9L9zc3AAA7777LkaPHg1XV1d069YNSUlJOHToEF5//fVCtW/WrFlo0aIFGjZsiPT0dGzbtk0OV0RUvjHcEJFZ7Nq1Cx4eHibH/P398ccffwB4tJIpPDwcr732Gjw8PPDtt9+iQYMGAABbW1vs3r0bEyZMwFNPPQVbW1v07t0bH3/8sVxXaGgo0tLSsGjRIrz11ltwdnZGnz59Ct0+KysrTJs2DVevXoWNjQ2efvpphIeHK/DJiai0SUIIYe5GEBEZkyQJmzdvRkhIiLmbQkQqxDk3REREVKEw3BAREVGFwjk3RFTucLSciEqCPTdERERUoTDcEBERUYXCcENEREQVCsMNERERVSgMN0RERFShMNwQERFRhcJwQ0RERBUKww0RERFVKAw3REREVKH8P8h8jpf/+otWAAAAAElFTkSuQmCC\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Iterations for Step: 1000\n",
            "Iterations for Sigmoid: 1000\n",
            "Iterations for ReLU: 1000\n"
          ]
        }
      ],
      "source": [
        "# Activation functions\n",
        "def step_activation(x):\n",
        "    return 1 if x >= 0 else 0\n",
        "\n",
        "def sigmoid_activation(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "def relu_activation(x):\n",
        "    return max(0, x)\n",
        "\n",
        "# Perceptron output calculation\n",
        "def perceptron_output(weights, inputs, activation_fn):\n",
        "    output = sum(w * x for w, x in zip(weights, inputs))\n",
        "    return activation_fn(output)\n",
        "\n",
        "# XOR gate training data\n",
        "training_data = [\n",
        "    (0, 0, 0),\n",
        "    (0, 1, 1),\n",
        "    (1, 0, 1),\n",
        "    (1, 1, 0)\n",
        "]\n",
        "\n",
        "def train_perceptron(initial_weights, learning_rate=0.05, convergence_error=0.002, max_epochs=1000, activation_fn=step_activation):\n",
        "    # Initialize weights\n",
        "    weights = initial_weights\n",
        "\n",
        "    # Training loop\n",
        "    epochs = 0\n",
        "    errors = []\n",
        "\n",
        "    while True:\n",
        "        epoch_error = 0\n",
        "        for x1, x2, target in training_data:\n",
        "            inputs = [1, x1, x2]  # Add bias term\n",
        "            output = perceptron_output(weights, inputs, activation_fn)\n",
        "            error = target - output\n",
        "            epoch_error += error ** 2\n",
        "\n",
        "            # Weight updates\n",
        "            for i in range(len(weights)):\n",
        "                weights[i] += learning_rate * error * inputs[i]\n",
        "\n",
        "        errors.append(epoch_error)\n",
        "        epochs += 1\n",
        "\n",
        "        if epoch_error <= convergence_error or epochs >= max_epochs:\n",
        "            break\n",
        "\n",
        "    return epochs, weights, errors\n",
        "\n",
        "# Initial weights\n",
        "initial_weights = [10, 0.2, -0.75]\n",
        "\n",
        "# Train the perceptron for XOR gate with different activation functions\n",
        "activation_functions = [step_activation, sigmoid_activation, relu_activation]\n",
        "activation_names = ['Step', 'Sigmoid', 'ReLU']\n",
        "iterations = []\n",
        "\n",
        "for activation_fn, activation_name in zip(activation_functions, activation_names):\n",
        "    num_epochs, final_weights, error_history = train_perceptron(initial_weights, activation_fn=activation_fn)\n",
        "    iterations.append(num_epochs)\n",
        "\n",
        "    # Plotting epochs against error values\n",
        "    plt.plot(range(num_epochs), error_history, label=activation_name)\n",
        "\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Error')\n",
        "plt.title('Error vs. Epochs for XOR Gate')\n",
        "plt.legend(activation_names)\n",
        "plt.show()\n",
        "\n",
        "# Compare the number of iterations taken to converge\n",
        "for i, activation_name in enumerate(activation_names):\n",
        "    print(f\"Iterations for {activation_name}: {iterations[i]}\")\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z9wnth_Cdmp7"
      },
      "source": [
        "A3 using XOR gate logic"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "o-QlhZMrdmp7",
        "outputId": "6d18c263-2aa0-484b-e16c-6a100343c5ef",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "def step_activation(x):\n",
        "    return 1 if x >= 0 else 0\n",
        "\n",
        "def train_perceptron(initial_weights, learningRate=0.05, activation_func=step_activation, convergence_error=0.002, max_epochs=1000):\n",
        "    W0, W1, W2 = initial_weights\n",
        "\n",
        "    def perceptron_output(x1, x2):\n",
        "        return W0 + W1 * x1 + W2 * x2\n",
        "\n",
        "    #and gate logic\n",
        "    training_data = [\n",
        "        (0, 0, 0),\n",
        "        (0, 1, 1),\n",
        "        (1, 0, 1),\n",
        "        (1, 1, 0)\n",
        "    ]\n",
        "\n",
        "    epochs = 0\n",
        "    errors = []\n",
        "\n",
        "    while True:\n",
        "        epoch_error = 0\n",
        "        for x1, x2, target in training_data:\n",
        "            output = perceptron_output(x1, x2)\n",
        "            error = target - activation_func(output)\n",
        "            epoch_error = epoch_error + error ** 2\n",
        "            W0 = W0 + learningRate * error\n",
        "            W1 = W1 + learningRate * error * x1\n",
        "            W2 = W2 + learningRate * error * x2\n",
        "\n",
        "        errors.append(epoch_error)\n",
        "        epochs += 1\n",
        "\n",
        "        if epoch_error <= convergence_error or epochs >= max_epochs:\n",
        "            break\n",
        "\n",
        "    return epochs, [W0, W1, W2], errors\n",
        "\n",
        "# Initial weights\n",
        "initial_weights = [10, 0.2, -0.75]\n",
        "\n",
        "# List of learning rates to test\n",
        "learning_rates = [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1]\n",
        "\n",
        "# List to store the number of epochs for each learning rate\n",
        "num_epochs_list = []\n",
        "\n",
        "# Train the perceptron with varying learning rates\n",
        "for learning_rate in learning_rates:\n",
        "    num_epochs, final_weights, errors = train_perceptron(initial_weights, learningRate=learning_rate)\n",
        "    num_epochs_list.append(num_epochs)\n",
        "\n",
        "# Plotting learning rates vs. number of epochs\n",
        "plt.plot(learning_rates, num_epochs_list, marker='o', linestyle='-')\n",
        "plt.xlabel('Learning Rate')\n",
        "plt.ylabel('Number of Epochs')\n",
        "plt.title('Number of Epochs vs. Learning Rate')\n",
        "plt.grid(True)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BgHl7VJhdmp9"
      },
      "source": [
        "A5.Use customer data provided below. Build a perceptron & learn to classify the transactions as high or low value as provided in the below table. Use sigmoid as the activation function. Initialize the weights & learning rate with your choice."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "id": "0KB-qhVpdmp9",
        "outputId": "dcf06802-4de7-4503-dcf9-9638bd26763a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "  Customer  Payment  Label  Predicted Label\n",
            "0      C_1      386      1                1\n",
            "1      C_2      289      1                1\n",
            "2      C_3      393      1                1\n",
            "3      C_4      110      0                0\n",
            "4      C_5      280      1                1\n",
            "5      C_6      167      0                0\n",
            "6      C_7      271      1                1\n",
            "7      C_8      274      1                1\n",
            "8      C_9      148      0                0\n",
            "9     C_10      198      0                0\n"
          ]
        }
      ],
      "source": [
        "# Sigmoid activation function\n",
        "def sigmoid(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "# Perceptron output calculation\n",
        "def perceptron_output(weights, inputs, bias):\n",
        "    weighted_sum = np.dot(weights, inputs) + bias\n",
        "    return sigmoid(weighted_sum)\n",
        "\n",
        "# Given customer data\n",
        "data = {\n",
        "    'Customer': ['C_1', 'C_2', 'C_3', 'C_4', 'C_5', 'C_6', 'C_7', 'C_8', 'C_9', 'C_10'],\n",
        "    'Candies': [20, 16, 27, 19, 24, 22, 15, 18, 21, 16],\n",
        "    'Mangoes': [6, 3, 6, 1, 4, 1, 4, 4, 1, 2],\n",
        "    'Milk Packets': [2, 6, 2, 2, 2, 5, 2, 2, 4, 4],\n",
        "    'Payment': [386, 289, 393, 110, 280, 167, 271, 274, 148, 198]\n",
        "}\n",
        "\n",
        "# Convert data to a DataFrame\n",
        "data = pd.DataFrame(data)\n",
        "\n",
        "# Labels: High (1) or Low (0) value based on Payment\n",
        "data['Label'] = [1 if payment >= 250 else 0 for payment in data['Payment']]\n",
        "\n",
        "# Features and labels as NumPy arrays\n",
        "features = data[['Candies', 'Mangoes', 'Milk Packets']].values.T  # Transpose the matrix\n",
        "labels = data['Label'].values\n",
        "\n",
        "# Initialize weights and bias with random values\n",
        "np.random.seed(0)  # For reproducibility\n",
        "num_features = features.shape[0]\n",
        "weights = np.random.rand(num_features)\n",
        "bias = np.random.rand()\n",
        "\n",
        "# Learning rate\n",
        "learning_rate = 0.01\n",
        "\n",
        "# Training loop\n",
        "num_epochs = 1000  # You can adjust this as needed\n",
        "for epoch in range(num_epochs):\n",
        "    total_error = 0\n",
        "    for i in range(len(features.T)):\n",
        "        # Forward pass\n",
        "        input_data = features[:, i]\n",
        "        predicted = perceptron_output(weights, input_data, bias)\n",
        "\n",
        "        # Calculate error\n",
        "        error = labels[i] - predicted\n",
        "        total_error += abs(error)\n",
        "\n",
        "        # Update weights and bias\n",
        "        weights += learning_rate * error * input_data\n",
        "        bias += learning_rate * error\n",
        "\n",
        "    # Check for convergence\n",
        "    if total_error == 0:\n",
        "        break\n",
        "\n",
        "# Evaluation\n",
        "predictions = [perceptron_output(weights, input_data, bias) for input_data in features.T]\n",
        "predicted_labels = [1 if prediction >= 0.5 else 0 for prediction in predictions]\n",
        "\n",
        "# Add predicted labels to the DataFrame\n",
        "data['Predicted Label'] = predicted_labels\n",
        "\n",
        "# Display the data with predicted labels\n",
        "print(data[['Customer', 'Payment', 'Label', 'Predicted Label']])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pwhZjEWldmp-"
      },
      "source": [
        "A6.Compare the results obtained from above perceptron learning to the ones obtained with matrix pseudo-inverse."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "id": "AcLFkS1wdmp_",
        "outputId": "ad70ebc4-457e-4090-a90f-f1b1dbf8c004",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Perceptron Accuracy (Train): 1.0\n",
            "Perceptron Accuracy (Test): 0.5\n",
            "Matrix Pseudo-Inverse Accuracy (Train): 1.0\n",
            "Matrix Pseudo-Inverse Accuracy (Test): 0.5\n",
            "Perceptron Accuracy (Train): 1.0\n",
            "Perceptron Accuracy (Test): 0.5\n",
            "Matrix Pseudo-Inverse Accuracy (Train): 1.0\n",
            "Matrix Pseudo-Inverse Accuracy (Test): 0.5\n"
          ]
        }
      ],
      "source": [
        "# Define the dataset\n",
        "data = {\n",
        "    'Customer': ['C_1', 'C_2', 'C_3', 'C_4', 'C_5', 'C_6', 'C_7', 'C_8', 'C_9', 'C_10'],\n",
        "    'Candies': [20, 16, 27, 19, 24, 22, 15, 18, 21, 16],\n",
        "    'Mangoes': [6, 3, 6, 1, 4, 1, 4, 4, 1, 2],\n",
        "    'Milk Packets': [2, 6, 2, 2, 2, 5, 2, 2, 4, 4],\n",
        "    'Payment': [386, 289, 393, 110, 280, 167, 271, 274, 148, 198]\n",
        "}\n",
        "\n",
        "# Convert data to a pandas DataFrame\n",
        "df = pd.DataFrame(data)\n",
        "\n",
        "# Convert data to a NumPy array\n",
        "X = df[['Candies', 'Mangoes', 'Milk Packets']].values\n",
        "y = np.array([1 if payment >= 250 else 0 for payment in df['Payment']])\n",
        "\n",
        "# Split the data into training and testing sets\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)\n",
        "\n",
        "# Train a perceptron model\n",
        "perceptron = Perceptron()\n",
        "perceptron.fit(X_train, y_train)\n",
        "\n",
        "# Predictions using perceptron\n",
        "y_train_pred_perceptron = perceptron.predict(X_train)\n",
        "y_test_pred_perceptron = perceptron.predict(X_test)\n",
        "\n",
        "# Train a model using matrix pseudo-inverse\n",
        "# Calculate weights using the pseudo-inverse\n",
        "weights_pseudo_inverse = np.linalg.pinv(X_train).dot(y_train)\n",
        "\n",
        "# Predictions using pseudo-inverse\n",
        "y_train_pred_pseudo_inverse = (X_train.dot(weights_pseudo_inverse) >= 0.5).astype(int)\n",
        "y_test_pred_pseudo_inverse = (X_test.dot(weights_pseudo_inverse) >= 0.5).astype(int)\n",
        "\n",
        "# Evaluate both models\n",
        "accuracy_perceptron_train = accuracy_score(y_train, y_train_pred_perceptron)\n",
        "accuracy_perceptron_test = accuracy_score(y_test, y_test_pred_perceptron)\n",
        "\n",
        "accuracy_pseudo_inverse_train = accuracy_score(y_train, y_train_pred_pseudo_inverse)\n",
        "accuracy_pseudo_inverse_test = accuracy_score(y_test, y_test_pred_pseudo_inverse)\n",
        "\n",
        "print(\"Perceptron Accuracy (Train):\", accuracy_perceptron_train)\n",
        "print(\"Perceptron Accuracy (Test):\", accuracy_perceptron_test)\n",
        "print(\"Matrix Pseudo-Inverse Accuracy (Train):\", accuracy_pseudo_inverse_train)\n",
        "print(\"Matrix Pseudo-Inverse Accuracy (Test):\", accuracy_pseudo_inverse_test)\n",
        "# Predictions using pseudo-inverse\n",
        "y_train_pred_pseudo_inverse = (X_train.dot(weights_pseudo_inverse) >= 0.5).astype(int)\n",
        "y_test_pred_pseudo_inverse = (X_test.dot(weights_pseudo_inverse) >= 0.5).astype(int)\n",
        "\n",
        "# Evaluate both models\n",
        "accuracy_perceptron_train = accuracy_score(y_train, y_train_pred_perceptron)\n",
        "accuracy_perceptron_test = accuracy_score(y_test, y_test_pred_perceptron)\n",
        "\n",
        "accuracy_pseudo_inverse_train = accuracy_score(y_train, y_train_pred_pseudo_inverse)\n",
        "accuracy_pseudo_inverse_test = accuracy_score(y_test, y_test_pred_pseudo_inverse)\n",
        "\n",
        "print(\"Perceptron Accuracy (Train):\", accuracy_perceptron_train)\n",
        "print(\"Perceptron Accuracy (Test):\", accuracy_perceptron_test)\n",
        "print(\"Matrix Pseudo-Inverse Accuracy (Train):\", accuracy_pseudo_inverse_train)\n",
        "print(\"Matrix Pseudo-Inverse Accuracy (Test):\", accuracy_pseudo_inverse_test)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YhVvhI2zdmqB"
      },
      "source": [
        "A7. Develop the belowNeural Network. Use learning rate (α) = 0.05 with a Sigmoid activation function. Learn the weights of the network using back-propagation algorithm to implement above provided AND gate logic."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "6apf4lHKdmqC",
        "outputId": "2c9d7c30-ef2c-4f8f-ae75-03db0ce6f07f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predictions:\n",
            "[[0.23327416]\n",
            " [0.29609331]\n",
            " [0.30368578]\n",
            " [0.36311458]]\n"
          ]
        }
      ],
      "source": [
        "# Sigmoid activation function\n",
        "def sigmoid(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "# Derivative of the sigmoid function\n",
        "def sigmoid_derivative(x):\n",
        "    return x * (1 - x)\n",
        "\n",
        "# AND gate training data\n",
        "X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])  #inputs of an AND gate\n",
        "y = np.array([[0], [0], [0], [1]])  #outputs of an AND gate\n",
        "\n",
        "# Neural network architecture [Given in the problem]\n",
        "input_nodes = 2\n",
        "hidden_nodes = 2\n",
        "output_nodes = 1\n",
        "\n",
        "# Initialize weights and biases with random values\n",
        "np.random.seed(0)   #random number generator in python\n",
        "weights_input_hidden = np.random.uniform(size=(input_nodes, hidden_nodes))\n",
        "biases_hidden = np.zeros((1, hidden_nodes))\n",
        "\n",
        "weights_hidden_output = np.random.uniform(size=(hidden_nodes, output_nodes))\n",
        "biases_output = np.zeros((1, output_nodes))\n",
        "\n",
        "# Learning rate\n",
        "learning_rate = 0.05\n",
        "\n",
        "# Maximum number of iterations\n",
        "max_iterations = 1000\n",
        "\n",
        "# Convergence error threshold\n",
        "convergence_error_threshold = 0.002\n",
        "\n",
        "# Training loop\n",
        "for iteration in range(max_iterations):\n",
        "    # Forward pass\n",
        "    hidden_layer_input = np.dot(X, weights_input_hidden) + biases_hidden\n",
        "    hidden_layer_output = sigmoid(hidden_layer_input)\n",
        "\n",
        "    output_layer_input = np.dot(hidden_layer_output, weights_hidden_output) + biases_output\n",
        "    output_layer_output = sigmoid(output_layer_input)\n",
        "\n",
        "    # Calculate error\n",
        "    error = y - output_layer_output\n",
        "\n",
        "    # Check for convergence\n",
        "    if np.mean(np.abs(error)) <= convergence_error_threshold:\n",
        "        print(f\"Converged after {iteration} iterations\")\n",
        "        break\n",
        "\n",
        "    # Backward pass\n",
        "    d_output = error * sigmoid_derivative(output_layer_output)\n",
        "    d_hidden = d_output.dot(weights_hidden_output.T) * sigmoid_derivative(hidden_layer_output)\n",
        "\n",
        "    # Update weights and biases\n",
        "    weights_hidden_output = weights_hidden_output + hidden_layer_output.T.dot(d_output) * learning_rate\n",
        "    weights_input_hidden = weights_input_hidden + X.T.dot(d_hidden) * learning_rate\n",
        "\n",
        "    biases_output = biases_output + np.sum(d_output, axis=0, keepdims=True) * learning_rate\n",
        "    biases_hidden = biases_hidden + np.sum(d_hidden, axis=0, keepdims=True) * learning_rate\n",
        "\n",
        "# Evaluate the trained network\n",
        "predictions = sigmoid(np.dot(sigmoid(np.dot(X, weights_input_hidden) + biases_hidden), weights_hidden_output) + biases_output)\n",
        "print(\"Predictions:\")\n",
        "print(predictions)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MVozp960dmqD"
      },
      "source": [
        "A8. Repeat the above A1 experiment for XOR Gate logic. Keep the learning rate & activation function same as A1."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "id": "IP5QymtPdmqE",
        "outputId": "ae63a419-a783-4066-ca9d-af8e48207dac",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predictions:\n",
            "[[0.47544663]\n",
            " [0.50272827]\n",
            " [0.50556324]\n",
            " [0.52649632]]\n"
          ]
        }
      ],
      "source": [
        "# Sigmoid activation function\n",
        "def sigmoid(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "# Derivative of the sigmoid function\n",
        "def sigmoid_derivative(x):\n",
        "    return x * (1 - x)\n",
        "\n",
        "# XOR gate training data\n",
        "X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])  #inputs of XOR gate\n",
        "y = np.array([[0], [1], [1], [0]])  #outputs of XOR gate\n",
        "\n",
        "# Neural network architecture [Given in the problem]\n",
        "input_nodes = 2\n",
        "hidden_nodes = 2\n",
        "output_nodes = 1\n",
        "\n",
        "# Initialize weights and biases with random values\n",
        "np.random.seed(0)   #random number generator in python\n",
        "weights_input_hidden = np.random.uniform(size=(input_nodes, hidden_nodes))\n",
        "biases_hidden = np.zeros((1, hidden_nodes))\n",
        "\n",
        "weights_hidden_output = np.random.uniform(size=(hidden_nodes, output_nodes))\n",
        "biases_output = np.zeros((1, output_nodes))\n",
        "\n",
        "# Learning rate\n",
        "learning_rate = 0.05\n",
        "\n",
        "# Maximum number of iterations\n",
        "max_iterations = 1000\n",
        "\n",
        "# Convergence error threshold\n",
        "convergence_error_threshold = 0.002\n",
        "\n",
        "# Training loop\n",
        "for iteration in range(max_iterations):\n",
        "    # Forward pass\n",
        "    hidden_layer_input = np.dot(X, weights_input_hidden) + biases_hidden\n",
        "    hidden_layer_output = sigmoid(hidden_layer_input)\n",
        "\n",
        "    output_layer_input = np.dot(hidden_layer_output, weights_hidden_output) + biases_output\n",
        "    output_layer_output = sigmoid(output_layer_input)\n",
        "\n",
        "    # Calculate error\n",
        "    error = y - output_layer_output\n",
        "\n",
        "    # Check for convergence\n",
        "    if np.mean(np.abs(error)) <= convergence_error_threshold:\n",
        "        print(f\"Converged after {iteration} iterations\")\n",
        "        break\n",
        "\n",
        "    # Backward pass\n",
        "    d_output = error * sigmoid_derivative(output_layer_output)\n",
        "    d_hidden = d_output.dot(weights_hidden_output.T) * sigmoid_derivative(hidden_layer_output)\n",
        "\n",
        "    # Update weights and biases\n",
        "    weights_hidden_output = weights_hidden_output + hidden_layer_output.T.dot(d_output) * learning_rate\n",
        "    weights_input_hidden = weights_input_hidden + X.T.dot(d_hidden) * learning_rate\n",
        "\n",
        "    biases_output = biases_output + np.sum(d_output, axis=0, keepdims=True) * learning_rate #keepdims: a parameter that you can pass to various functions to control whether the dimensions of the result should be kept or reduced.\n",
        "    biases_hidden = biases_hidden + np.sum(d_hidden, axis=0, keepdims=True) * learning_rate\n",
        "\n",
        "# Evaluate the trained network\n",
        "predictions = sigmoid(np.dot(sigmoid(np.dot(X, weights_input_hidden) + biases_hidden), weights_hidden_output) + biases_output)\n",
        "print(\"Predictions:\")\n",
        "print(predictions)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9qaiikqAdmqG"
      },
      "source": [
        "A9. Repeat exercise A1 & A2 with 2 output nodes (as shown below). A zero output of logic gate maps to [O1O2] = [1 0] from output layer while a one output from logic gate maps to [0 1]."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NOp65KyYdmqH"
      },
      "source": [
        "Using AND gate"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "0HVJMj_xdmqH",
        "outputId": "a93d9472-1f37-454e-ff0b-6d0671b06a4c",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predictions:\n",
            "[[0.73341184 0.23289031]\n",
            " [0.72086012 0.2929836 ]\n",
            " [0.71829714 0.30085794]\n",
            " [0.70764963 0.35833452]]\n"
          ]
        }
      ],
      "source": [
        "# Sigmoid activation function\n",
        "def sigmoid(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "# Derivative of the sigmoid function\n",
        "def sigmoid_derivative(x):\n",
        "    return x * (1 - x)\n",
        "\n",
        "# AND gate training data\n",
        "X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])  #inputs of AND gate\n",
        "y = np.array([[1, 0], [1, 0], [1, 0], [0, 1]])   #outputs of AND gate\n",
        "\n",
        "# Neural network architecture [Given in the problem]\n",
        "input_nodes = 2\n",
        "hidden_nodes = 2\n",
        "output_nodes = 2\n",
        "\n",
        "# Initialize weights and biases with random values\n",
        "np.random.seed(0)   #random number generator in python\n",
        "weights_input_hidden = np.random.uniform(size=(input_nodes, hidden_nodes))\n",
        "biases_hidden = np.zeros((1, hidden_nodes))\n",
        "\n",
        "weights_hidden_output = np.random.uniform(size=(hidden_nodes, output_nodes))\n",
        "biases_output = np.zeros((1, output_nodes))\n",
        "\n",
        "# Learning rate\n",
        "learning_rate = 0.05\n",
        "\n",
        "# Maximum number of iterations\n",
        "max_iterations = 1000\n",
        "\n",
        "# Convergence error threshold\n",
        "convergence_error_threshold = 0.002\n",
        "\n",
        "# Training loop\n",
        "for iteration in range(max_iterations):\n",
        "    # Forward pass\n",
        "    hidden_layer_input = np.dot(X, weights_input_hidden) + biases_hidden\n",
        "    hidden_layer_output = sigmoid(hidden_layer_input)\n",
        "\n",
        "    output_layer_input = np.dot(hidden_layer_output, weights_hidden_output) + biases_output\n",
        "    output_layer_output = sigmoid(output_layer_input)\n",
        "\n",
        "    # Calculate error\n",
        "    error = y - output_layer_output\n",
        "\n",
        "    # Check for convergence\n",
        "    if np.mean(np.abs(error)) <= convergence_error_threshold:\n",
        "        print(f\"Converged after {iteration} iterations\")\n",
        "        break\n",
        "\n",
        "    # Backward pass\n",
        "    d_output = error * sigmoid_derivative(output_layer_output)\n",
        "    d_hidden = d_output.dot(weights_hidden_output.T) * sigmoid_derivative(hidden_layer_output)\n",
        "\n",
        "    # Update weights and biases\n",
        "    weights_hidden_output = weights_hidden_output + hidden_layer_output.T.dot(d_output) * learning_rate\n",
        "    weights_input_hidden = weights_input_hidden + X.T.dot(d_hidden) * learning_rate\n",
        "\n",
        "    biases_output = biases_output + np.sum(d_output, axis=0, keepdims=True) * learning_rate #keepdims: a parameter that you can pass to various functions to control whether the dimensions of the result should be kept or reduced.\n",
        "    biases_hidden = biases_hidden + np.sum(d_hidden, axis=0, keepdims=True) * learning_rate\n",
        "\n",
        "# Evaluate the trained network\n",
        "predictions = sigmoid(np.dot(sigmoid(np.dot(X, weights_input_hidden) + biases_hidden), weights_hidden_output) + biases_output)\n",
        "print(\"Predictions:\")\n",
        "print(predictions)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7M_e7ysWdmqI"
      },
      "source": [
        "Using XOR gate"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "VN8uSPundmqJ",
        "outputId": "3857d19e-cb16-453a-f03d-113c71dc4b0c",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Predictions:\n",
            "[[0.48741224 0.46977702]\n",
            " [0.49773916 0.50128228]\n",
            " [0.49821744 0.50484499]\n",
            " [0.50697855 0.5310057 ]]\n"
          ]
        }
      ],
      "source": [
        "# Sigmoid activation function\n",
        "def sigmoid(x):\n",
        "    return 1 / (1 + np.exp(-x))\n",
        "\n",
        "# Derivative of the sigmoid function\n",
        "def sigmoid_derivative(x):\n",
        "    return x * (1 - x)\n",
        "\n",
        "# XOR gate training data\n",
        "X = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])  #inputs of XOR gate\n",
        "y = np.array([[1, 0], [0, 1], [0, 1], [1, 0]])   #outputs of XOR gate\n",
        "\n",
        "# Neural network architecture [Given in the problem]\n",
        "input_nodes = 2\n",
        "hidden_nodes = 2\n",
        "output_nodes = 2\n",
        "\n",
        "# Initialize weights and biases with random values\n",
        "np.random.seed(0)   #random number generator in python\n",
        "weights_input_hidden = np.random.uniform(size=(input_nodes, hidden_nodes))\n",
        "biases_hidden = np.zeros((1, hidden_nodes))\n",
        "\n",
        "weights_hidden_output = np.random.uniform(size=(hidden_nodes, output_nodes))\n",
        "biases_output = np.zeros((1, output_nodes))\n",
        "\n",
        "# Learning rate\n",
        "learning_rate = 0.05\n",
        "\n",
        "# Maximum number of iterations\n",
        "max_iterations = 1000\n",
        "\n",
        "# Convergence error threshold\n",
        "convergence_error_threshold = 0.002\n",
        "\n",
        "# Training loop\n",
        "for iteration in range(max_iterations):\n",
        "    # Forward pass\n",
        "    hidden_layer_input = np.dot(X, weights_input_hidden) + biases_hidden\n",
        "    hidden_layer_output = sigmoid(hidden_layer_input)\n",
        "\n",
        "    output_layer_input = np.dot(hidden_layer_output, weights_hidden_output) + biases_output\n",
        "    output_layer_output = sigmoid(output_layer_input)\n",
        "\n",
        "    # Calculate error\n",
        "    error = y - output_layer_output\n",
        "\n",
        "    # Check for convergence\n",
        "    if np.mean(np.abs(error)) <= convergence_error_threshold:\n",
        "        print(f\"Converged after {iteration} iterations\")\n",
        "        break\n",
        "\n",
        "    # Backward pass\n",
        "    d_output = error * sigmoid_derivative(output_layer_output)\n",
        "    d_hidden = d_output.dot(weights_hidden_output.T) * sigmoid_derivative(hidden_layer_output)\n",
        "\n",
        "    # Update weights and biases\n",
        "    weights_hidden_output = weights_hidden_output + hidden_layer_output.T.dot(d_output) * learning_rate\n",
        "    weights_input_hidden = weights_input_hidden + X.T.dot(d_hidden) * learning_rate\n",
        "\n",
        "    biases_output = biases_output + np.sum(d_output, axis=0, keepdims=True) * learning_rate #keepdims: a parameter that you can pass to various functions to control whether the dimensions of the result should be kept or reduced.\n",
        "    biases_hidden = biases_hidden + np.sum(d_hidden, axis=0, keepdims=True) * learning_rate\n",
        "\n",
        "# Evaluate the trained network\n",
        "predictions = sigmoid(np.dot(sigmoid(np.dot(X, weights_input_hidden) + biases_hidden), weights_hidden_output) + biases_output)\n",
        "print(\"Predictions:\")\n",
        "print(predictions)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9LrRhGO6dmqK"
      },
      "source": [
        "A10.  Repeat the AND Gate and XOR Gate exercises using MLPClassifier() function."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k0QQ6LvNdmqK"
      },
      "source": [
        "Using AND gate logic"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "Fou9kKnbdmqL",
        "outputId": "ca85cb23-5717-423d-849b-ee2be28e94a4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "AND gate Predictions: \n",
            "[0 0 0 1]\n"
          ]
        }
      ],
      "source": [
        "#defining AND gate training data\n",
        "X = np.array([[0,0],[0,1],[1,0],[1,1]])\n",
        "y = np.array([0,0,0,1])\n",
        "\n",
        "#creating MLPClassifier\n",
        "classiAND = MLPClassifier(hidden_layer_sizes=(2,), activation='logistic', solver='lbfgs',max_iter=1000)\n",
        "\n",
        "#training the classifier\n",
        "classiAND.fit(X,y)\n",
        "\n",
        "#test the classifier\n",
        "testData = np.array([[0,0],[0,1],[1,0],[1,1]])\n",
        "predictions = classiAND.predict(testData)\n",
        "\n",
        "#printing the result\n",
        "print(\"AND gate Predictions: \")\n",
        "print(predictions)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WGlGbQJodmqL"
      },
      "source": [
        "Using XOR gate logic"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "id": "Y33R-oKhdmqM",
        "outputId": "94e71de2-ae18-45ec-8979-db250b500762",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "XOR gate Predictions:\n",
            "[0 1 1 0]\n"
          ]
        }
      ],
      "source": [
        "# defining XOR gate training data\n",
        "X_xor = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])\n",
        "y_xor = np.array([0, 1, 1, 0])\n",
        "\n",
        "# creating MLPClassifier\n",
        "classiXOR = MLPClassifier(hidden_layer_sizes=(4, 4), activation='logistic', solver='lbfgs', max_iter=1000)\n",
        "\n",
        "# training the classifier\n",
        "classiXOR.fit(X_xor, y_xor)\n",
        "\n",
        "# test the classifier\n",
        "testData_xor = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])\n",
        "predictions_xor = classiXOR.predict(testData_xor)\n",
        "\n",
        "# printing the result\n",
        "print(\"XOR gate Predictions:\")\n",
        "print(predictions_xor)"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.10.5"
    },
    "orig_nbformat": 4,
    "colab": {
      "provenance": []
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}