siamese_network.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "siamese network.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lm22QzrJDMgu"
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.layers import Input, Flatten, Dense, Dropout, Lambda\n",
        "from tensorflow.keras.optimizers import RMSprop\n",
        "from tensorflow.keras.datasets import fashion_mnist\n",
        "from tensorflow.python.keras.utils.vis_utils import plot_model\n",
        "from tensorflow.keras import backend as K\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from PIL import Image, ImageFont, ImageDraw\n",
        "import random"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9ODSqMJJVWLd"
      },
      "source": [
        "def create_pairs(x, digit_indices):\n",
        "\n",
        "    pairs = []\n",
        "    labels = []\n",
        "    n = min([len(digit_indices[d]) for d in range(10)]) - 1\n",
        "    \n",
        "    for d in range(10):\n",
        "        for i in range(n):\n",
        "            z1, z2 = digit_indices[d][i], digit_indices[d][i + 1]\n",
        "            pairs += [[x[z1], x[z2]]]\n",
        "            inc = random.randrange(1, 10)\n",
        "            dn = (d + inc) % 10\n",
        "            z1, z2 = digit_indices[d][i], digit_indices[dn][i]\n",
        "            pairs += [[x[z1], x[z2]]]\n",
        "            labels += [1, 0]\n",
        "            \n",
        "    return np.array(pairs), np.array(labels)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eRFKYmK9Xnx6"
      },
      "source": [
        "def create_pairs_on_set(images, labels):\n",
        "    \n",
        "    digit_indices = [np.where(labels == i)[0] for i in range(10)]\n",
        "    pairs, y = create_pairs(images, digit_indices)\n",
        "    y = y.astype('float32')\n",
        "    \n",
        "    return pairs, y"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fCqRUvI_YAls"
      },
      "source": [
        "def show_image(image):\n",
        "  plt.figure()\n",
        "  plt.imshow(image)\n",
        "  plt.colorbar()\n",
        "  plt.grid(False)\n",
        "  plt.show()\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IQAqfvmfYiOC"
      },
      "source": [
        "(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()\n",
        "train_images = train_images.astype('float32')\n",
        "test_images = test_images.astype('float32')\n",
        "train_images = train_images / 255.0\n",
        "test_images = test_images / 255.0\n",
        "tr_pairs, tr_y = create_pairs_on_set(train_images, train_labels)\n",
        "ts_pairs, ts_y = create_pairs_on_set(test_images, test_labels)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 538
        },
        "id": "kGzIzwPyYzmm",
        "outputId": "bf316d22-54e4-4e7d-e63c-ba7aca2faa31"
      },
      "source": [
        "i=5\n",
        "\n",
        "show_image(tr_pairs[i][0])\n",
        "show_image(ts_pairs[i][0])\n",
        "print(tr_y[i])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "0.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AvCQJ74SZgbx"
      },
      "source": [
        "def base_network():\n",
        "  input = Input(shape=(28,28,), name='base_input')\n",
        "  x = Flatten(name='Flatten_input')(input)\n",
        "  x = Dense(128, activation='relu', name = 'f1')(x)\n",
        "  x = Dropout(0.1, name = 'D1')(x)\n",
        "  x = Dense(128, activation='relu', name = 'f2')(x)\n",
        "  x = Dropout(0.1, name = 'D2')(x)\n",
        "  x = Dense(128, activation='relu', name = 'f3')(x)\n",
        "\n",
        "  return Model(inputs = input, outputs =  x)\n",
        "\n",
        "def euclidean_distance(vects):\n",
        "    x, y = vects\n",
        "    sum_square = K.sum(K.square(x - y), axis=1, keepdims=True)\n",
        "    return K.sqrt(K.maximum(sum_square, K.epsilon()))\n",
        "\n",
        "def euclidian_distance_shape(shapes):\n",
        "  s1, s2 = shapes\n",
        "  return (s1[0], 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DZ0Grsz_gyIs"
      },
      "source": [
        "base_network = base_network()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 661
        },
        "id": "SWWc21CVg6zA",
        "outputId": "c8500bf5-52b0-41ff-a3c3-cb4c046f1037"
      },
      "source": [
        "plot_model(base_network, show_layer_names=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ggChflA5iLeh"
      },
      "source": [
        "*italicized text*"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bE1VLQWGiPVb"
      },
      "source": [
        "input1 = Input(shape=(28, 28,), name = \"i1\")\n",
        "v1 = base_network(input1)\n",
        "\n",
        "input2 = Input(shape = (28, 28, ), name = 'i2')\n",
        "v2 = base_network(input2)\n",
        "\n",
        "output = Lambda(euclidean_distance, name = 'output_layer', output_shape = euclidian_distance_shape)([v1, v2])\n",
        "\n",
        "model = Model(inputs = [input1, input2], outputs = output)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 272
        },
        "id": "Sc_dvTENj-cp",
        "outputId": "7d04606f-7d7a-4e77-8a22-5d8589ab7647"
      },
      "source": [
        "plot_model(model, show_layer_names=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAASsAAAD/CAYAAACpfPA5AAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nO3dd1gUd/4H8Pcsu+yyCyygCFhAAY0ajVExKpbTmHqeORURLDH6aA7LiV6wxBKDLQliyQXQBDWeYqNGsefkNHaCseBZsMWCRkF6Uxb4/P7w3F82ItJnZ/m8noc/mJ2deS/75c3ud4ZZgYgIjDFm3KJlYidgjLHK4LJijEkClxVjTBK4rBhjkiD/44KTJ09i5cqVYmRhRiw6OrpOtsvjjZWnvPH23Curu3fvIiYmpl4CMeOXmppap+OBxxv7vYrG23OvrJ6pq7+kTFqioqLg4+NT5/vh8caAiscbz1kxxiSBy4oxJglcVowxSeCyYoxJApcVY0wSuKwYY5LAZcUYkwQuK8aYJHBZMcYkgcuKMSYJXFaMMUngsmKMSQKXFWNMErisGGOSUCtltXfvXmi1Wuzateu528rKyrBq1Sp4enpWe/unTp1Cu3btIJPJIAgCHBwcsGTJkppErnWxsbFwdXWFIAgQBAGOjo4YPXq02LFMUnnjbdGiRWjfvj2sra2hVCrh7u6OWbNmIT8/v8rb5/FmpOgPIiMjqZzFFdq9ezdZW1tTfHy8wfKrV69Sr169CAB16tSpStssz7vvvksAKCsrq8bbqitubm6k1WrFjlFrqjMe6nr75Y23P/3pTxQWFkYZGRmUm5tLkZGRpFAo6L333qt2Nh5v9a+C8RD1wovvVcXAgQORk5NjsOz8+fNYtGgRJk2ahIKCApCJfTxhUVERBgwYgBMnTogdpcEpb7xZWlrCz88PZmZmAIDhw4cjNjYWUVFRuHv3Llq0aCFG1FrD460O56w6deqE2NhYjBo1Ckqlsq52I5r169cjLS1N7Bjsf3bv3q0vqmcaN24MACgsLBQjUq3i8VYLZXXs2DE4OztDEASEhoZW+f779++HtbU1li5dWuX7rl69GhqNBmq1Gjt37sT7778Pa2trNG/eHNu2bdOv980330ClUqFJkyaYOHEinJycoFKp4OnpicTERP16/v7+MDc3h6Ojo37ZlClToNFoIAgCHj16BACYPn06AgICcOPGDQiCAHd39ypnB4CjR4+iffv20Gq1UKlU6NixIw4cOAAAmDBhgn4+ws3NDWfPngUAjBs3Dmq1GlqtFvHx8QCA0tJSLFiwAM7OzrCwsMBrr72GyMhIAMCyZcugVqthZWWFtLQ0BAQEoFmzZkhJSalWZrFVZbzdu3cPFhYWaNWqlX4ZjzcJj7cqvGd8obt37xIACgkJKff27t27v3DOavfu3WRlZUWLFi166X7Km0OYN28eAaCEhATKycmhtLQ06tOnD2k0GiouLtav5+fnRxqNhi5dukSPHz+mixcvUrdu3cjKyoru3LmjX2/UqFHk4OBgsN/g4GACQOnp6fplXl5e5Obm9lzGqswhREdHU2BgIGVmZlJGRgb16NGDGjVqZLAPMzMzunfvnsH9Ro4caTBfM2PGDFIqlRQTE0NZWVk0d+5ckslklJSUZPAzmjZtGoWEhNDQoUPp8uXLlcpojHNWLxtvREQFBQVkZWVF/v7+Bst5vEl2vEWJXlZVUdHgKSoq0i8LCwsjAHT9+nX9Mj8/v+ee1KSkJAJACxcu1C+rz8HzR1988QUBoLS0NCIiOnjwIAGgJUuW6NfJycmh1q1bU0lJCRERFRUVkVqtJl9fX/06hYWFpFQqafLkyURU/s+osqRaVvPmzaM2bdpQbm5utbPxeDOq8RZlkudZmZubAwB0Ol2F63l4eECtVuPKlSv1EeulFAoFgKcvswHgzTffRJs2bfD999/rD1Bs374dvr6++vmZlJQUFBYWokOHDvrtWFhYwNHR0WgeV32Li4tDVFQUDhw4ACsrqzrfH4+3+hlvJllWVaFUKpGeni7Kvvfs2YN+/frB3t4eSqUSs2bNMrhdEARMnDgRN2/eREJCAgBg06ZNGD9+vH6dgoICAMD8+fP1cw6CIOD27dsmMbFcVdu3b8dXX32Fw4cPo2XLlmLHeQ6Pt+pr0GWl0+mQnZ2N5s2b18v+jhw5glWrVgEA7ty5gyFDhsDR0RGJiYnIyclBUFDQc/cZO3YsVCoV1q1bh5SUFFhbW8PFxUV/u729PQBg1apVICKDr5MnT9bL4zIWISEh2Lx5M/7zn/+gadOmYsd5Do+3mqmV86yk6vDhwyAi9OjRQ79MLpe/9OV8df3yyy/QaDQAgAsXLkCn02Hy5MlwdXUF8PQv2x/Z2trCx8cH27dvh5WVFT7++GOD21u0aAGVSoVz587VSWYpICJ8+umnyMrKwo4dOyCXG+ew5vFWM6K/stq3b1+1DyVXVVlZGbKyslBSUoLk5GRMnz4dzs7OGDt2rH4dd3d3ZGZmYseOHdDpdEhPT8ft27ef25adnR3u37+PW7duIS8vr8IBp9Pp8PDhQxw+fFg/eJydnQEABw8exOPHj3Ht2jWDw9q/N2nSJDx58gS7d+/GoEGDDG5TqVQYN24ctm3bhtWrVyM3NxelpaVITU3Fb7/9VtUfkSRdunQJy5Ytw9q1a6FQKAzengiCgOXLl+vX5fEm4fFWhdn4coWEhJCjoyMBILVaTR988AEREZ08eZJ69epFTk5OBIAAkKOjI3l6etJPP/2kv//evXvJysrK4AjEH506dYpeffVVkslk+u0sXbqUwsLCSK1WEwBq3bo13bhxg8LDw8na2poAkIuLC129epWInh6dUSgU1KxZM5LL5WRtbU2DBw+mGzduGOwrIyOD+vfvTyqVilq1akVTp06lmTNnEgByd3fXH3Y+c+YMubi4kIWFBfXu3ZvWrFlDbm5u+sf6oq+4uDj9vmbPnk12dnZkY2ND3t7eFBoaSgDIzc3N4PA2EVHnzp1pzpw55f58njx5QrNnzyZnZ2eSy+Vkb29PXl5edPHiRQoKCiILCwsCQC1atKCIiIhKP7dExnc0sLzxduHChQp/5sHBwfr783iT7HirnVMXpMDPz4/s7OzEjlFtf/7zn+nmzZv1vl9jKyup4PFWPQ3u1IUXeXaIVgp+/zI/OTkZKpXK4ExsZvx4vNUu45yJZJg9ezYmTZoEIsK4ceMQEREhdiRmwqQw3hrEK6u5c+diw4YNyMnJQatWrRATEyN2pJdSq9Vo27Yt3nrrLQQGBqJ9+/ZiR2KVxOOtbghEhtduiYqKgo+Pj8ld0oVVT12PBx5v7PcqGA/RDeKVFWNM+risGGOSwGXFGJMELivGmCRwWTHGJIHLijEmCVxWjDFJ4LJijEkClxVjTBK4rBhjksBlxRiTBC4rxpgkcFkxxiThhdez8vb2rs8cJu/x48dQKBT6z1+TitTU1HrZjxTHW0FBgf4a56x2VDTenntl1aJFCwwbNqxOAzVER48exenTp8WOUWXNmzev0/Eg1fGWmpqK/fv3Izc3V+woJqWi8fbc9axY3UhISMB7772HoKAgfPLJJ2LHYTWQnJwMT09PjBs3DiEhIWLHaSiiuazq0bJlyzB37lzs3bsX77zzjthxWDVkZmbijTfegJOTExISEvQfHc/qHJdVfSIi+Pr6IiEhAUlJSUZ3QX5WsdLSUvzlL3/BpUuXkJSUhCZNmogdqSHhsqpv+fn56NmzJ+RyOY4fPw61Wi12JFZJ06dPR3h4OI4cOQIPDw+x4zQ0fFnj+mZpaYldu3bh7t278PPzEzsOq6SIiAh88803WL9+PReVSLisRNCyZUts3boV27ZtQ1hYmNhx2EucOXMGfn5+mDVrFkaMGCF2nAaL3waKaPHixVi8eDEOHjyIvn37ih2HlePhw4fw8PBAu3btsG/fPsmdJ2dCeM5KTEQEb29vHDt2DKdPn0bz5s3FjsR+R6fT4a233sLdu3eRlJSERo0aiR2pIeM5KzEJgoANGzagUaNG8Pb2xpMnT8SOxH7n73//O86cOYP4+HguKiPAZSUyKysrxMXF4fLly5g2bZrYcdj/rFmzBmvXrsWGDRvQoUMHseMwcFkZhVdeeQUbN25EeHg41q5dK3acBu/EiROYPn06AgMDJfmvQKaK56yMyPz587Fs2TIcOnQIvXr1EjtOg3T//n14eHjgjTfeQFxcHGQy/ntuJHiC3ZiUlZVh0KBBOHv2LE6fPo2mTZuKHalBefz4Mfr27Yu8vDwkJibC2tpa7Ejs//EEuzGRyWTYvHkzNBoNvL29UVxcLHakBmXy5Mm4fv06du3axUVlhLisjIytrS3i4uJw/vx5zJw5U+w4Dcby5cuxceNGbNmyBe7u7mLHYeXgsjJCHTt2xKZNmxASEoINGzaIHcfkHTx4EHPmzEFQUBDef/99seOwF+A5KyM2c+ZMhIaG4ujRo/z/aHXk1q1b6NatG9555x1s2bJF7DjsxXiC3ZiVlpZi4MCBuHz5Mk6fPg17e3uxI5mU/Px8eHp6wszMjK+AYfx4gt2YmZmZYevWrZDL5fD19UVJSYnYkUwGEWH8+PG4f/8+4uLiuKgkgMvKyNnZ2SEuLg6nTp3C3LlzxY5jMhYvXowffvgBMTExfBFEieCykoBOnTohPDwcy5cvR2RkpNhxJC8+Ph4LFy7EP//5T/Tr10/sOKySeM5KQvz9/bFu3TocP34cnTt3FjuOJKWkpKB79+4YPHgw/vWvf4kdh1UeT7BLSUlJCd566y3cvn0bp0+f5isBVFFubi569OgBGxsbHDp0CEqlUuxIrPJ4gl1K5HI5IiMjUVJSghEjRqC0tFTsSJJRVlaGkSNHIjs7G9HR0VxUEsRlJTEODg6IiYnBkSNHEBgYKHYcyZg7dy4SEhLwww8/oFmzZmLHYdXAZSVB3bt3R3h4OJYuXYqYmBix4xi9uLg4LFu2DKGhoejevbvYcVg18ZyVhPn5+WHr1q04deoUXn31VbHjGKXz58+jV69emDBhAr7++mux47Dq4wl2KdPpdHjzzTeRlpaGn3/+GVqtVuxIRiUjIwNvvPEGmjVrhoSEBCgUCrEjserjCXYpUygUiI6ORn5+PsaMGYOysjKxIxmNkpISeHt7o7S0FLGxsVxUJoDLSuIcHR0RExOD/fv3Y+nSpWLHMRr/+Mc/cOrUKcTGxvL/VJoILisT0LNnT6xYsQKBgYHYs2eP2HFEt2nTJoSFhWHDhg3o2rWr2HFYLeE5KxMyfvx4xMXFISkpqcFeQO7UqVPo168fAgIC+JWmaeEJdlPS0K8h/uDBA3Tr1g3t27fH3r17+dOTTQtPsJsSlUqFmJgYZGRkYOzYsSjv7xARobCwUIR0tePx48flLtfpdBg+fDjUajUiIyO5qEwQl5WJcXZ2RmRkJHbt2oXly5cb3Jabm4shQ4YgPDxcpHQ1N2HCBHz55ZfPFfGUKVNw7tw5xMXFwcbGRqR0rE4RM0nLly8nmUxG+/btIyKiy5cvk5ubGwGg119/XeR01ZOXl0cqlYoAkJeXF+Xn5xMRUUhICMlkMoqPjxc5IatDUVxWJuyjjz4iOzs7Wrt2LVlaWpJcLicABICuXr0qdrwq27RpE8lkMgJAcrmc2rVrR1FRUWRubk5LliwROx6rW1E8wW7C8vPz0bdvX5w7dw6CIOhPGlUoFPj8888xb948kRNWzdtvv41Dhw7przahUCigVCrRv39/7Ny5E4IgiJyQ1SGeYDdVubm58PX1xfnz50FEBme363Q6yX3EV1pamkFRAU8fR1FREfbs2YNly5aJmI7VBy4rE3TlyhV07twZP/744wv/BefGjRu4cOFCPServq1bt5b7yqm0tBRlZWWYM2cORowYgaKiIhHSsfrAZWVijh49ii5duuDXX3+FTqd74Xrm5ubYtm1bPSarmY0bN1Z4sUEiwvbt29GvXz9kZmbWYzJWX7isTEyvXr0QFBQECwsLmJubv3C94uJibNq0qdxzsYxNSkoKzp07V2FWuVwOOzs7+Pv7w87Orh7TsfrCZWViZDIZpk6diuvXr2PgwIH6ZeW5d+8eEhMT6zNetWzZsuWFV00wMzODIAjw9fXF1atXMWrUqHpOx+oLl5WJcnJyQlxcHOLj49GkSRPI5fLn1pHKW8FNmzaV+5ZWJpOhTZs2OH78OCIiIvgDNEwcl5WJGzRoEK5fv46AgADIZDKDf0MpLi5GRESEUX/S88mTJ3H79m2DZc9OWfjss89w7tw59OzZU6R0rD5xWTUAGo0GX331FZKSkvDqq68aFFZWVhYOHz4sXriX+P1bwGdvZwcMGICUlBQEBgZWOC/HTAuXVQPSpUsXnD59GosXL4a5uTkUCgVkMpnRvhXU6XTYsmULdDod5HI5HBwcsGPHDuzbtw8uLi5ix2P1TDJnsKempuLEiRNixzAZDx8+xHfffYeLFy/CwsICa9euNbpL/545cwZBQUEQBAEDBw7E8OHD+fP+akmLFi2k9vZZOtezioqKgo+Pj9gxGDMJw4YNQ3R0tNgxqiL6+UNERk4i3Sopjx49wvHjx/HXv/5V7Ch6paWl2Lp1K0aNGvXCUy9Y9Xh7e4sdoVokV1as9jVu3Nioigp4ev7Uhx9+KHYMZkT4TxZjTBK4rBhjksBlxRiTBC4rxpgkcFkxxiSBy4oxJglcVowxSeCyYoxJApcVY0wSuKwYY5LAZcUYkwQuK8aYJHBZMcYkweTL6smTJ5g2bRocHR2hVquxf/9+sSMBAPbu3QutVotdu3aJsv/ly5ejSZMmEAQB3377bbW3U1ZWhlWrVsHT07Pa24iNjYWrqysEQXjhV8uWLau9/bpgKs+flJh8Wa1YsQL79+/HlStX8PXXXyM/P1/sSADEvy7XjBkzanzl1WvXrqFv37745JNPUFhYWO3teHl54ebNm3Bzc4NWqwURgYhQUlKCwsJCPHz4EGq1ukZZa5spPH9SY/LXs9qxYwc8PDxgY2ODv/3tb6JkKCoqwoABAwwG18CBA5GTkyNKntpw/vx5LFq0CJMmTUJBQUGd/PKamZnBwsICFhYWaNOmTa1vv7JM8fmTIpN/ZZWamir6tcXXr1+PtLQ0UTPUtk6dOiE2NhajRo2ql+ui79ixo8738SKm+PxJkcmW1b///W+4u7vjt99+w8aNGyEIAiwtLeHv7w9zc3M4Ojrq150yZQo0Gg0EQcCjR48AAKtXr4ZGo4FarcbOnTvx/vvvw9raGs2bNy/302AiIiLg4eEBlUoFjUaDli1bYvHixZg+fToCAgJw48YNCIIAd3d3HDt2DM7OzhAEAaGhofptEBFWrlyJdu3aQalUwtbWFoMHD8aVK1f061Ql19GjR9G+fXtotVqoVCp07NgRBw4cqO0fdYX2798Pa2trLF26tFa2x89f/T5/RoUkIjIykqoT18HBgT766CODZaNGjSIHBweDZcHBwQSA0tPT9cvmzZtHACghIYFycnIoLS2N+vTpQxqNhoqLi/XrrVq1igDQl19+SRkZGZSZmUnfffcdjRo1ioiIvLy8yM3NzWB/d+/eJQAUEhKiX7ZgwQIyNzeniIgIys7OpuTkZOrSpQs1btyYHjx4UOVc0dHRFBgYSJmZmZSRkUE9evSgRo0a6W+/du0aAaA1a9ZU+ef6e927d6dOnTqVe9vu3bvJysqKFi1a9NLtuLm5kVarNViWkJBAwcHBBsv4+Xuqus/fsGHDaNiwYVW6jxGIMtlXVrXJ09MT1tbWsLe3h6+vLwoKCnDnzh0ATz/bbuHChejfvz8+/fRT2NnZwdbWFuPHj0e3bt0qvY+ioiKsXLkSQ4cOxejRo6HVatGxY0d8++23ePToEcLDw6uUC3j6CSaff/45bG1tYWdnhw8++AAZGRlIT0+v+Q+lkgYOHIjc3Fx89tlnlVo/JyfH4CjggAEDapyBnz/TwGVVRc8+AVin0wEAkpOTkZ2djXfffddgPTMzM0ybNq3S27148SLy8/Ph4eFhsLxbt24wNzdHYmJilXKV59ncXWlpaaVz1bffHw0kIhw6dKhWt8/Pn3SZ/NHAupabmwsAsLGxqdF2srOzAQCWlpbP3WZjY4O8vLwqb3PPnj0IDg7GxYsXkZubW+EvgrHq168f+vXrV2fb5+dPOviVVQ01bdoUAPQTu9X17JelvEGdnZ2N5s2bV2l7d+7cwZAhQ+Do6IjExETk5OQgKCioRhlNET9/0tEgy0oul9faX6mWLVvCzs4OP/74Y42206FDB1haWuL06dMGyxMTE1FcXIyuXbtWaXsXLlyATqfD5MmT4erqCpVKBUEQapTRWPDz1zA1yLJyd3dHZmYmduzYAZ1Oh/T0dNy+fbta21IqlZg7dy6OHDkCf39/3Lt3D2VlZcjLy8OlS5cAAHZ2drh//z5u3bqFvLy8cn/RVCoVAgICEBcXh82bNyM3NxcXLlzApEmT4OTkBD8/vyrlcnZ2BgAcPHgQjx8/xrVr1146b1IX9u3bV6unLgD8/DVYIh+OrLSqnrpw69Yt6ty5MwEguVxOXbp0oZiYGCIiysjIoP79+5NKpaJWrVrR1KlTaebMmQSA3N3d6c6dOxQWFkZqtZoAUOvWrenGjRsUHh5O1tbWBIBcXFzo6tWr+v2FhoZSx44dSaVSkUqlos6dO1NYWBgREZ05c4ZcXFzIwsKCevfuTfPnzydHR0cCQGq1mj744AMiIiorK6Pg4GBq3bo1KRQKsrW1pSFDhlBKSop+P1XJNXv2bLKzsyMbGxvy9vam0NBQAkBubm40ffp0cnBwIACk0Who6NChVXo+Tp48Sb169SInJycCQADI0dGRPD096aefftKvt3fvXrKysqIlS5a8cFvHjx+nNm3aGGxnwIABL1yfn7+aPX9SPXVBIBL5n5wqKSoqCj4+PqL/TxZjUuft7Q0AiI6OFjlJlUQ3yLeBjDHp4bJiAIArV65UeImWZ1++vr5iR2UNFJ9nxQAAbdu25bfYzKjxKyvGmCRwWTHGJIHLijEmCVxWjDFJ4LJijEkClxVjTBK4rBhjksBlxRiTBC4rxpgkcFkxxiSBy4oxJglcVowxSeCyYoxJApcVY0wSJHeJmKioKLEjMCZpqampVf60HWMgubLy8fEROwJjkjds2DCxI1SZZK7BzqRBEARERkZi+PDhYkdhpoWvwc4YkwYuK8aYJHBZMcYkgcuKMSYJXFaMMUngsmKMSQKXFWNMErisGGOSwGXFGJMELivGmCRwWTHGJIHLijEmCVxWjDFJ4LJijEkClxVjTBK4rBhjksBlxRiTBC4rxpgkcFkxxiSBy4oxJglcVowxSeCyYoxJApcVY0wSuKwYY5LAZcUYkwQuK8aYJHBZMcYkgcuKMSYJXFaMMUngsmKMSQKXFWNMErisGGOSwGXFGJMEudgBmHSFh4cjKyvrueU7d+7Er7/+arBs7NixcHBwqK9ozAQJRERih2DS5Ofnh/DwcCiVSv0yIoIgCPrvS0pKoNVq8eDBAygUCjFiMtMQzW8DWbWNGDECAPDkyRP9V3FxscH3MpkMI0aM4KJiNcZlxaqtb9++aNKkSYXr6HQ6fakxVhNcVqzaZDIZRo8eDXNz8xeu4+TkBE9Pz3pMxUwVlxWrkREjRqC4uLjc2xQKBcaMGWMwh8VYdXFZsRrx8PBAq1atyr2N3wKy2sRlxWpszJgx5U6gu7q6olOnTiIkYqaIy4rV2OjRo6HT6QyWKRQKjBs3TqREzBRxWbEac3d3R8eOHQ3mpnQ6HXx8fERMxUwNlxWrFWPGjIGZmRkAQBAEdO7cGa1btxY5FTMlXFasVowcORKlpaUAADMzM3z00UciJ2KmhsuK1YqmTZvC09MTgiCgrKwM3t7eYkdiJobLitWaDz/8EESEvn37omnTpmLHYSbG5P6RmU9AZA1dZGQkhg8fLnaM2hZtkpeImT59Onr27Cl2jAZpxYoV8PPzg6WlpdhRGiRTPgJrkmXVs2dPU/zLIgmenp5o3ry52DEaLFMuK56zYrWKi4rVFS4rxpgkcFkxxiSBy4oxJglcVowxSeCyYoxJApcVY0wSuKwYY5LAZcUYkwQuK8aYJHBZMcYkgcuKMSYJXFaMMUngsmKMSQKXVQOyfPlyNGnSBIIg4NtvvxU7TpXExsbC1dUVgiBAEAQ4Ojpi9OjRomTp1q0bzMzM8Prrr1f5vhMmTICVlRUEQcC5c+fqIJ3p4rJqQGbMmIETJ06IHaNavLy8cPPmTbi5uUGr1eLBgwfYvHmzKFmSkpLQv3//at133bp1WLt2bS0nahi4rGpZUVERPD09Jb8P9nJ8Ce36xWVVy9avX4+0tDTJ74O9nEKhqNb9uOSqp8GXFRFh5cqVaNeuHZRKJWxtbTF48GBcuXJFv46/vz/Mzc3h6OioXzZlyhRoNBoIgoBHjx4BeHrt94CAANy4cQOCIMDd3R3ffPMNVCoVmjRpgokTJ8LJyQkqlQqenp5ITEyslX3U1NGjR9G+fXtotVqoVCp07NgRBw4cAPB0juXZPJGbmxvOnj0LABg3bhzUajW0Wi3i4+MBAKWlpViwYAGcnZ1hYWGB1157DZGRkQCAZcuWQa1Ww8rKCmlpaQgICECzZs2QkpKC/fv3w9raGkuXLq3xY6ns4/r666+h0Wggk8nQtWtXODg4QKFQQKPRoEuXLujTpw9atGgBlUoFGxsbzJo167ntX79+HW3btoVGo4GFhQX69OmDY8eOGaxDRAgODsYrr7wCpVIJrVaLmTNnVikr+x8yMQAoMjKy0usvWLCAzM3NKSIigrKzsyk5OZm6dOlCjRs3pgcPHujXGzVqFDk4OBjcNzg4mABQenq6fpmXlxe5ubkZrOfn50cajYYuXbpEjx8/posXL1K3bt3IysqK7ty5Uyv7qKxr164RAFqzZo1+WXR0NAUGBlJmZiZlZGRQjx49qFGjRgb7MzMzo3v37hlsa+TIkRQfH6//fsaMGaRUKikmJoaysrJo7ty5JJPJKCkpiYiI5s2bRwBo2rRpFBISQkOHDqXLly/T7t27ycrKihYtWvTS/G5ubqTVaiv1WF/2uD7//HMCQHZUe1YAAAbgSURBVImJiVRQUECPHj2i9957jwDQnj17KD09nQoKCsjf358A0Llz5/T3HTBgALm6utKvv/5KOp2O/vvf/1L37t1JpVLR1atX9evNmzePBEGgFStWUFZWFhUWFlJYWBgBoLNnz1Y6a2VVdfxLSFSDLqvCwkKytLQkX19fg+U///wzATD45alpWf3xFywpKYkA0MKFC2tlH5VVXln90RdffEEAKC0tjYiIDh48SABoyZIl+nVycnKodevWVFJSQkRERUVFpFarDX6WhYWFpFQqafLkyUT0/2VVVFRUrexEVSurP/rj43pWVnl5efp1Nm7cSADowoUL+mXPxsP27dv1ywYMGECdOnUy2H5ycjIBoBkzZhDR08evVqvp7bffNlhv27Ztz5XVy7JWlimXVYN+G3jx4kXk5+fDw8PDYHm3bt1gbm5u8Dattnl4eECtVhu83TQWz+Zinn0c/Jtvvok2bdrg+++/B/3vYya3b98OX19fmJmZAQBSUlJQWFiIDh066LdjYWEBR0dHo3mMf3xc5TE3NwcAlJSUPHc/nU5X4fY7duwIrVaL5ORkAE/fJhYWFmLAgAF1krWhadBllZ2dDQDlfsadjY0N8vLy6nT/SqUS6enpdbqPytizZw/69esHe3t7KJXK5+ZnBEHAxIkTcfPmTSQkJAAANm3ahPHjx+vXKSgoAADMnz9fP8clCAJu376NwsLC+nswv/Oyx1UXFAqFvtRSU1MBAPb29i+9nxhZpaZBl5WNjQ0AlFtK2dnZdfqxUjqdrs73URl37tzBkCFD4OjoiMTEROTk5CAoKOi59caOHQuVSoV169YhJSUF1tbWcHFx0d/+7Bdy1apVICKDr5MnT9bLYzly5AhWrVpVpcdVm0pKSpCZmQlnZ2cAgEqlAgA8efKkwvuJkVWKTPJDTiurQ4cOsLS0xOnTpw2WJyYmori4GF27dtUvk8vlL30bUBWHDx8GEaFHjx51to/KuHDhAnQ6HSZPngxXV1cA5R9at7W1hY+PD7Zv3w4rKyt8/PHHBrc/O3Im5lnZv/zyCzQaDYDKP67adOjQIZSVlaFLly4Ano4vmUyGn376CZMmTXrh/cTIKkUN+pWVSqVCQEAA4uLisHnzZuTm5uLChQuYNGkSnJyc4Ofnp1/X3d0dmZmZ2LFjB3Q6HdLT03H79u3ntmlnZ4f79+/j1q1byMvL05dPWVkZsrKyUFJSguTkZEyfPh3Ozs4YO3Zsre2jOp69Cjh48CAeP36Ma9euvXCubtKkSXjy5Al2796NQYMGGdymUqkwbtw4bNu2DatXr0Zubi5KS0uRmpqK3377rcIM+/btq9GpCzqdDg8fPsThw4f1ZVWVx1VdxcXFyMnJQUlJCc6cOQN/f3+4uLjon1N7e3t4eXkhJiYG69evR25uLpKTkxEeHm6wnfrIahLEnN6vC6ji0ZCysjIKDg6m1q1bk0KhIFtbWxoyZAilpKQYrJeRkUH9+/cnlUpFrVq1oqlTp9LMmTMJALm7u+tPQThz5gy5uLiQhYUF9e7dmx48eEB+fn6kUCioWbNmJJfLydramgYPHkw3btyotX1UxooVK8jBwYEAkEajoaFDhxIR0ezZs8nOzo5sbGzI29ubQkNDCQC5ubkZnFpBRNS5c2eaM2dOudt/8uQJzZ49m5ydnUkul5O9vT15eXnRxYsXKSgoiCwsLAgAtWjRgiIiIvT327t3L1lZWRkcbfyjuLg4cnNzIwAVfsXFxenvU9HjCggIILVaTQCoZcuWdPToUfrqq69Iq9USAHJwcKAtW7bQ9u3b9T8zW1tb2rZtGxERbdiwgfr3709NmjQhuVxOjRo1ohEjRtDt27cNcufl5dGECROoUaNGZGlpSb1796YFCxYQAGrevDmdP3++ys9BRao6/iUkSiD63+EdEyEIAiIjIzF8+HCxo+hNnDgR0dHRyMjIEDtKjQ0cOBChoaFo1aqV2FFYOYxx/NeS6Ab9NrA+SfUQ9O/fYiYnJ0OlUnFRMVFwWUnclStXDE4VeNGXr69vtbY/e/ZsXLt2DVevXsW4ceOwePHiWn4EjFVOgz4aWB/mzp2LDRs2oLi4GK1atUJwcDCGDRtWa9tv27Yt6vKdvFqtRtu2bdGsWTOEhYWhffv2dbYvxirCc1aMmRATHv88Z8UYkwYuK8aYJHBZMcYkgcuKMSYJXFaMMUngsmKMSQKXFWNMErisGGOSwGXFGJMELivGmCRwWTHGJIHLijEmCVxWjDFJMMlLxPj4+MDHx0fsGIyxWmRyZRUZGSl2BMZE5enpKXaEOmFy17NijJkkvp4VY0wauKwYY5LAZcUYkwQ5gGixQzDG2Euc+j9YNVGM7weKzgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XpvBilULkChV"
      },
      "source": [
        "def contrastive_loss_with_margin(margin):\n",
        "    def contrastive_loss(y_true, y_pred):\n",
        "\n",
        "        square_pred = K.square(y_pred)\n",
        "        margin_square = K.square(K.maximum(margin - y_pred, 0))\n",
        "        return K.mean(y_true * square_pred + (1 - y_true) * margin_square)\n",
        "    return contrastive_loss"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "S9nk88Vxvxdv",
        "outputId": "181f6a8e-b1ba-4288-a0ee-3a273a563745"
      },
      "source": [
        "rms = RMSprop()\n",
        "model.compile(loss = contrastive_loss_with_margin(margin=1), optimizer=rms)\n",
        "history = model.fit([tr_pairs[:,0], tr_pairs[:,1]], tr_y, epochs=20, batch_size=128, validation_data=([ts_pairs[:,0], ts_pairs[:,1]], ts_y))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "938/938 [==============================] - 9s 10ms/step - loss: 0.1137 - val_loss: 0.0849\n",
            "Epoch 2/20\n",
            "938/938 [==============================] - 7s 7ms/step - loss: 0.0813 - val_loss: 0.0739\n",
            "Epoch 3/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0721 - val_loss: 0.0727\n",
            "Epoch 4/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0673 - val_loss: 0.0669\n",
            "Epoch 5/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0641 - val_loss: 0.0685\n",
            "Epoch 6/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0621 - val_loss: 0.0664\n",
            "Epoch 7/20\n",
            "938/938 [==============================] - 8s 8ms/step - loss: 0.0609 - val_loss: 0.0647\n",
            "Epoch 8/20\n",
            "938/938 [==============================] - 8s 8ms/step - loss: 0.0592 - val_loss: 0.0642\n",
            "Epoch 9/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0581 - val_loss: 0.0677\n",
            "Epoch 10/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0573 - val_loss: 0.0642\n",
            "Epoch 11/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0562 - val_loss: 0.0649\n",
            "Epoch 12/20\n",
            "938/938 [==============================] - 8s 8ms/step - loss: 0.0556 - val_loss: 0.0646\n",
            "Epoch 13/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0547 - val_loss: 0.0642\n",
            "Epoch 14/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0544 - val_loss: 0.0634\n",
            "Epoch 15/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0540 - val_loss: 0.0634\n",
            "Epoch 16/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0532 - val_loss: 0.0636\n",
            "Epoch 17/20\n",
            "938/938 [==============================] - 8s 8ms/step - loss: 0.0527 - val_loss: 0.0642\n",
            "Epoch 18/20\n",
            "938/938 [==============================] - 8s 8ms/step - loss: 0.0522 - val_loss: 0.0662\n",
            "Epoch 19/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0519 - val_loss: 0.0640\n",
            "Epoch 20/20\n",
            "938/938 [==============================] - 7s 8ms/step - loss: 0.0514 - val_loss: 0.0657\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8QjvOki9wcxj"
      },
      "source": [
        "def compute_accuracy(y_true, y_pred):\n",
        "    '''Compute classification accuracy with a fixed threshold on distances.\n",
        "    '''\n",
        "    pred = y_pred.ravel() > 0.5\n",
        "    return np.mean(pred == y_true)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qusu6edXxCv-",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c823a8cc-a9f5-4c9d-9c77-72005c2cd753"
      },
      "source": [
        "loss = model.evaluate(x=[ts_pairs[:,0],ts_pairs[:,1]], y=ts_y)\n",
        "\n",
        "y_pred_train = model.predict([tr_pairs[:,0], tr_pairs[:,1]])\n",
        "train_accuracy = compute_accuracy(tr_y, y_pred_train)\n",
        "\n",
        "y_pred_test = model.predict([ts_pairs[:,0], ts_pairs[:,1]])\n",
        "test_accuracy = compute_accuracy(ts_y, y_pred_test)\n",
        "\n",
        "print(\"Loss = {}, Train Accuracy = {} Test Accuracy = {}\".format(loss, train_accuracy, test_accuracy))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "625/625 [==============================] - 1s 1ms/step - loss: 0.0657\n",
            "Loss = 0.0656726136803627, Train Accuracy = 0.06433572262043674 Test Accuracy = 0.0916916916916917\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SwwJCO-cxEME"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}