3D_MNIST-TF2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "252301a7-63a4-4380-93a6-69aec0b2b272",
   "metadata": {},
   "source": [
    "# Conv-3D: 3-D MNIST with TensorFlow2.8\n",
    "\n",
    "You can refer to 3D MNIST dataset [here](https://www.kaggle.com/daavoo/3d-mnist#).\n",
    "\n",
    "Example codes can be referenced [here](https://www.kaggle.com/shivamb/3d-convolutions-understanding-use-case) & [here](https://github.com/christianversloot/machine-learning-articles/blob/main/a-simple-conv3d-example-with-keras.md)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bfe8415d-c066-47a8-95db-6a71d198449b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specify GPU to be used-\n",
    "%env CUDA_DEVICE_ORDER=PCI_BUS_ID\n",
    "%env CUDA_VISIBLE_DEVICES = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "51606994-509b-4bc5-86b7-d2830eab5565",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "3cef97e8-7610-4e14-8ab7-1b05eb6858ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Flatten, Conv3D, MaxPooling3D\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from tensorflow.keras.models import Sequential, Model\n",
    "\n",
    "import tensorflow as tf\n",
    "import h5py\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8960663a-cc01-4219-97eb-2032356867e7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9fe510cb-a878-4923-b489-067d2a819bad",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a4ab0a46-7b68-4509-9eff-51b046e59ca0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TensorFlow version: 2.8.0\n"
     ]
    }
   ],
   "source": [
    "print(f\"TensorFlow version: {tf.__version__}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a7461f8e-3eb1-44df-8b04-03236fd9e62b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of GPUs available = 0\n",
      "No GPU found\n"
     ]
    }
   ],
   "source": [
    "num_gpus = len(tf.config.list_physical_devices('GPU'))\n",
    "print(f\"number of GPUs available = {num_gpus}\")\n",
    "\n",
    "# Check GPU availibility-\n",
    "gpu_devices = tf.config.list_physical_devices('GPU')\n",
    "\n",
    "if gpu_devices:\n",
    "    print(f\"GPU: {gpu_devices}\")\n",
    "    details = tf.config.experimental.get_device_details(gpu_devices[0])\n",
    "    print(f\"GPU details: {details.get('device_name', 'Unknown GPU')}\")\n",
    "else:\n",
    "    print(\"No GPU found\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7385dea5-64a9-427f-be0d-0864a2e64ca6",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1eb482c1-5558-49b5-ba02-a2caf371e48a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "bfb23d6b-1456-40ac-8d3f-cc5823348376",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specify hyper-parameters-\n",
    "batch_size = 64\n",
    "no_epochs = 35\n",
    "learning_rate = 0.001\n",
    "num_classes = 10\n",
    "validation_split = 0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ca33a065-0ae8-4410-8833-4d08bb1a6e88",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "790bcb2e-3dc2-44b6-9a21-66b80518eaff",
   "metadata": {},
   "source": [
    "#### Define some helper function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "dcb84d6a-47ae-4b19-909b-c1fd3e639f16",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert 1D vector into 3D values, provided by the 3D MNIST authors at\n",
    "# https://www.kaggle.com/daavoo/3d-mnist\n",
    "def array_to_color(array, cmap=\"Oranges\"):\n",
    "    '''\n",
    "    The imported data has only 1 channel. This function converts the data into 3 channel - RGB format.\n",
    "    This ensures resemblence with the original 2D scenario.\n",
    "    '''\n",
    "    s_m = plt.cm.ScalarMappable(cmap=cmap)\n",
    "    return s_m.to_rgba(array)[:,:-1]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fd757e77-dd94-4cb4-9cce-c8059cf8faf0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Reshape data into format that can be handled by Conv3D layers.\n",
    "# Courtesy of Sam Berglin; Zheming Lian; Jiahui Jang - University of Wisconsin-Madison\n",
    "# Report - https://github.com/sberglin/Projects-and-Papers/blob/master/3D%20CNN/Report.pdf\n",
    "# Code - https://github.com/sberglin/Projects-and-Papers/blob/master/3D%20CNN/network_final_version.ipynb\n",
    "def rgb_data_transform(data):\n",
    "    '''\n",
    "    This function reshapes the data, which per sample comes in as 1-D shape: (4096,)  - 16x16x16 pixels = 4096 pixels,\n",
    "    which is a 1-D array. This function reshapes the data into three-channeled, four-dimensional 16x16x16x3 format by\n",
    "    making use of 'array_to_color()'. The Conv3D function can now handle the data.\n",
    "    '''\n",
    "    data_t = []\n",
    "    for i in range(data.shape[0]):\n",
    "        data_t.append(array_to_color(data[i]).reshape(16, 16, 16, 3))\n",
    "    return np.asarray(data_t, dtype=np.float32)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9c4fd533-39ad-420e-8c19-c87559a1652d",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e0ed815d-ce45-42c2-b30c-3cbb437a0034",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "07506dc8-5f26-491f-b4d5-f0f89cda63f1",
   "metadata": {},
   "source": [
    "### 3-D MNIST data preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "39e21fa0-0480-487c-82aa-f3d9671b0e28",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Arjun\\AppData\\Local\\Temp\\ipykernel_8504\\1228886426.py:20: DeprecationWarning: Converting `np.integer` or `np.signedinteger` to a dtype is deprecated. The current result is `np.dtype(np.int_)` which is not strictly correct. Note that the result depends on the system. To ensure stable results use may want to use `np.int64` or `np.int32`.\n",
      "  y_train = to_categorical(y_train).astype(np.integer)\n",
      "C:\\Users\\Arjun\\AppData\\Local\\Temp\\ipykernel_8504\\1228886426.py:21: DeprecationWarning: Converting `np.integer` or `np.signedinteger` to a dtype is deprecated. The current result is `np.dtype(np.int_)` which is not strictly correct. Note that the result depends on the system. To ensure stable results use may want to use `np.int64` or `np.int32`.\n",
      "  y_test = to_categorical(y_test).astype(np.integer)\n"
     ]
    }
   ],
   "source": [
    "# Load the HDF5 data file\n",
    "with h5py.File(\"3D_MNIST-archive/full_dataset_vectors.h5\", \"r\") as hf:\n",
    "\n",
    "    # Split the data into training, validation features & targets/labels-\n",
    "    X_train = hf[\"X_train\"][:]\n",
    "    y_train = hf[\"y_train\"][:]\n",
    "    X_test = hf[\"X_test\"][:] \n",
    "    y_test = hf[\"y_test\"][:]\n",
    "\n",
    "    # Determine sample shape-\n",
    "    sample_shape = (16, 16, 16, 3)\n",
    "\n",
    "    # Reshape data into 3D format-\n",
    "    # Transform & reshape data from one-channeled (4096,) format into three-channeled\n",
    "    # (16, 16, 16, 3) format.\n",
    "    X_train = rgb_data_transform(X_train)\n",
    "    X_test = rgb_data_transform(X_test)\n",
    "\n",
    "    # Convert target vectors to categorical targets (format)-\n",
    "    y_train = to_categorical(y_train).astype(np.integer)\n",
    "    y_test = to_categorical(y_test).astype(np.integer)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "05490fee-d1b8-4a69-99b7-f4ddcfaee039",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((10000, 16, 16, 16, 3), (2000, 16, 16, 16, 3))"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Sanity check-\n",
    "X_train.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "0ba4abd7-aa77-479a-a7ff-dfd517697de2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((10000, 10), (2000, 10))"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Sanity check-\n",
    "y_train.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1e13f977-1e78-478d-92a8-447a4d4e5087",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "d50ffad3-3c26-4507-ab6b-78ac2a3c9614",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Maximum pixel value = 1.000 & minimum pixel value = 0.004\n"
     ]
    }
   ],
   "source": [
    "print(f\"Maximum pixel value = {X_train.max():.3f} & minimum pixel value = {X_train.min():.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cfb11aec-b997-4f5e-9a4b-8ab9d609c275",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f9e5a984-5ef0-4872-ab23-7c70ca9acb3c",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "a2123f28-9b97-4aab-99e5-0a703c704f2d",
   "metadata": {},
   "source": [
    "### Define Conv3D architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "82b06dd6-09fa-41d2-b957-340662f7b162",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Conv3D_Architecture(Model):\n",
    "    def __init__(self):\n",
    "        super(Conv3D_Architecture, self).__init__()\n",
    "        \n",
    "        self.conv1 = Conv3D(\n",
    "            filters = 32, kernel_size = (3, 3, 3),\n",
    "            activation = None, kernel_initializer = 'he_uniform',\n",
    "            input_shape = sample_shape\n",
    "        )\n",
    "        \n",
    "        self.pool1 = MaxPooling3D(\n",
    "            pool_size = (2, 2, 2)\n",
    "        )\n",
    "        \n",
    "        self.conv2 = Conv3D(\n",
    "            filters = 64, kernel_size = (3, 3, 3),\n",
    "            activation = None, kernel_initializer = 'he_uniform'\n",
    "        )\n",
    "        '''\n",
    "        self.pool2 = MaxPooling3D(\n",
    "            pool_size = (2, 2, 2)\n",
    "        )\n",
    "        '''\n",
    "        self.flatten = Flatten()\n",
    "        \n",
    "        self.dense1 = Dense(\n",
    "            units = 256, activation = None,\n",
    "            kernel_initializer = 'he_uniform'\n",
    "        )\n",
    "        \n",
    "        self.output_layer = Dense(\n",
    "            units = num_classes, activation = None\n",
    "        )\n",
    "        \n",
    "        \n",
    "    def call(self, x):\n",
    "        x = tf.nn.relu(self.conv1(x))\n",
    "        x = self.pool1(x)\n",
    "        x = tf.nn.relu(self.conv2(x))\n",
    "        x = self.pool1(x)\n",
    "        x = self.flatten(x)\n",
    "        x = tf.nn.relu(self.dense1(x))\n",
    "        x = tf.nn.softmax(self.output_layer(x))\n",
    "        return x\n",
    "    \n",
    "    \n",
    "    def shape_computation(self, x):\n",
    "        print(f\"Input shape: {x.shape}\")\n",
    "        x = self.conv1(x)\n",
    "        print(f\"first conv3D output shape: {x.shape}\")\n",
    "        x = self.pool1(x)\n",
    "        print(f\"first MaxPooling3D output shape: {x.shape}\")\n",
    "        x = self.conv2(x)\n",
    "        print(f\"second conv3D output shape: {x.shape}\")\n",
    "        x = self.pool1(x)\n",
    "        print(f\"second MaxPooling3D output shape: {x.shape}\")\n",
    "        x = self.flatten(x)\n",
    "        print(f\"Flattened output shape: {x.shape}\")\n",
    "        x = self.dense1(x)\n",
    "        print(f\"first dense output shape: {x.shape}\")\n",
    "        x = self.output_layer(x)\n",
    "        print(f\"Output layer shape: {x.shape}\")\n",
    "        del x\n",
    "        return None\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b6ffbd7d-c94b-4d99-83ba-566004d965f2",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "2ff16217-7163-406a-9cff-78c8ea8ef2c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Initialize an instance of 3D CNN architecture-\n",
    "model = Conv3D_Architecture()\n",
    "model.build(input_shape = (None, 16, 16, 16, 3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4f9c101b-7104-4177-81d6-09c7948667b0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "ddacddca-796c-4e08-baa8-06c181ffb418",
   "metadata": {},
   "outputs": [],
   "source": [
    "x = X_train[:5, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "b9e9e396-115f-4f81-bdb5-8d81cc60a9db",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_out = model(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "ad5e622a-5267-4b9e-9594-d9b55ceb9ceb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Input shape: (5, 16, 16, 16, 3)\n",
      "first conv3D output shape: (5, 14, 14, 14, 32)\n",
      "first MaxPooling3D output shape: (5, 7, 7, 7, 32)\n",
      "second conv3D output shape: (5, 5, 5, 5, 64)\n",
      "second MaxPooling3D output shape: (5, 2, 2, 2, 64)\n",
      "Flattened output shape: (5, 512)\n",
      "first dense output shape: (5, 256)\n",
      "Output layer shape: (5, 10)\n"
     ]
    }
   ],
   "source": [
    "model.shape_computation(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "caa410af-87e7-47fd-9a24-5e405d671546",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(TensorShape([5, 10]), (5, 16, 16, 16, 3))"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_out.shape, x.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "1b66fd72-ddad-4138-bc85-77b2a0082b54",
   "metadata": {},
   "outputs": [],
   "source": [
    "del x_out, x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a89dd86f-f66f-4643-8455-c05acc7a5e17",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "43724191-611c-44e8-8bd2-84e2ac31845c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"conv3d__architecture_3\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv3d_6 (Conv3D)           multiple                  2624      \n",
      "                                                                 \n",
      " max_pooling3d_3 (MaxPooling  multiple                 0         \n",
      " 3D)                                                             \n",
      "                                                                 \n",
      " conv3d_7 (Conv3D)           multiple                  55360     \n",
      "                                                                 \n",
      " flatten_3 (Flatten)         multiple                  0         \n",
      "                                                                 \n",
      " dense_6 (Dense)             multiple                  131328    \n",
      "                                                                 \n",
      " dense_7 (Dense)             multiple                  2570      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 191,882\n",
      "Trainable params: 191,882\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Print model summary-\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1d5bb690-269c-4d2c-bcdd-7c6c7d4db858",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "e7856fb5-f315-421d-8906-fa27d6d69fe3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "layer: (3, 3, 3, 3, 32) has 2592 trainable params\n",
      "layer: (32,) has 0 trainable params\n",
      "layer: (3, 3, 3, 32, 64) has 55296 trainable params\n",
      "layer: (64,) has 0 trainable params\n",
      "layer: (512, 256) has 131072 trainable params\n",
      "layer: (256,) has 0 trainable params\n",
      "layer: (256, 10) has 2560 trainable params\n",
      "layer: (10,) has 0 trainable params\n"
     ]
    }
   ],
   "source": [
    "# Get layer-wise trainable parameters-\n",
    "tot_params = 0\n",
    "\n",
    "for layer in model.trainable_weights:\n",
    "    loc_params = tf.math.count_nonzero(layer, axis = None).numpy()\n",
    "    print(f\"layer: {layer.shape} has {loc_params} trainable params\")\n",
    "    tot_params += loc_params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "c7709f76-722f-42a7-bab1-a71ef7d18e92",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of trainable parameters = 191520\n"
     ]
    }
   ],
   "source": [
    "print(f\"Total number of trainable parameters = {tot_params}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4ccc9064-657b-490d-a3b9-7d2a1a4a7bdd",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "2e352410-0fa6-4275-b9bc-fdb20480a9aa",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compile define model-\n",
    "model.compile(\n",
    "    loss = tf.keras.losses.categorical_crossentropy,\n",
    "    optimizer = tensorflow.keras.optimizers.Adam(learning_rate = learning_rate),\n",
    "    metrics = ['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b8cdc53a-2ec4-40e2-93a8-011879c16671",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "021cd5c9-ce7d-4074-882d-65e1d2c0cada",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define early stopping criterion-\n",
    "early_stopping = tf.keras.callbacks.EarlyStopping(\n",
    "    monitor = 'val_loss', min_delta = 0.001,\n",
    "    patience = 5, restore_best_weights = True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d3314b93-2549-4a32-a56b-3aa4d287b79f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "da583ab2-4746-4cf8-8bb5-9cb9d0e1903a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "dcfa568d-7391-4e5f-8f52-31010c9466e2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/35\n",
      "125/125 [==============================] - 42s 331ms/step - loss: 2.1867 - accuracy: 0.2246 - val_loss: 1.8990 - val_accuracy: 0.3320\n",
      "Epoch 2/35\n",
      "125/125 [==============================] - 41s 330ms/step - loss: 1.6715 - accuracy: 0.4232 - val_loss: 1.5184 - val_accuracy: 0.4845\n",
      "Epoch 3/35\n",
      "125/125 [==============================] - 42s 336ms/step - loss: 1.4386 - accuracy: 0.5065 - val_loss: 1.3990 - val_accuracy: 0.5355\n",
      "Epoch 4/35\n",
      "125/125 [==============================] - 40s 322ms/step - loss: 1.3121 - accuracy: 0.5483 - val_loss: 1.2921 - val_accuracy: 0.5640\n",
      "Epoch 5/35\n",
      "125/125 [==============================] - 41s 328ms/step - loss: 1.2334 - accuracy: 0.5730 - val_loss: 1.2291 - val_accuracy: 0.5895\n",
      "Epoch 6/35\n",
      "125/125 [==============================] - 41s 329ms/step - loss: 1.1695 - accuracy: 0.5897 - val_loss: 1.2397 - val_accuracy: 0.5915\n",
      "Epoch 7/35\n",
      "125/125 [==============================] - 41s 331ms/step - loss: 1.1217 - accuracy: 0.6096 - val_loss: 1.1606 - val_accuracy: 0.6100\n",
      "Epoch 8/35\n",
      "125/125 [==============================] - 42s 332ms/step - loss: 1.0889 - accuracy: 0.6210 - val_loss: 1.2360 - val_accuracy: 0.5825\n",
      "Epoch 9/35\n",
      "125/125 [==============================] - 41s 331ms/step - loss: 1.0528 - accuracy: 0.6326 - val_loss: 1.1250 - val_accuracy: 0.6225\n",
      "Epoch 10/35\n",
      "125/125 [==============================] - 42s 335ms/step - loss: 1.0131 - accuracy: 0.6460 - val_loss: 1.1392 - val_accuracy: 0.6135\n",
      "Epoch 11/35\n",
      "125/125 [==============================] - 41s 326ms/step - loss: 0.9883 - accuracy: 0.6503 - val_loss: 1.1418 - val_accuracy: 0.6205\n",
      "Epoch 12/35\n",
      "125/125 [==============================] - 42s 333ms/step - loss: 0.9657 - accuracy: 0.6670 - val_loss: 1.1086 - val_accuracy: 0.6300\n",
      "Epoch 13/35\n",
      "125/125 [==============================] - 41s 327ms/step - loss: 0.9297 - accuracy: 0.6790 - val_loss: 1.1120 - val_accuracy: 0.6250\n",
      "Epoch 14/35\n",
      "125/125 [==============================] - 41s 328ms/step - loss: 0.9045 - accuracy: 0.6904 - val_loss: 1.0712 - val_accuracy: 0.6395\n",
      "Epoch 15/35\n",
      "125/125 [==============================] - 42s 337ms/step - loss: 0.8995 - accuracy: 0.6914 - val_loss: 1.0695 - val_accuracy: 0.6450\n",
      "Epoch 16/35\n",
      "125/125 [==============================] - 43s 341ms/step - loss: 0.8533 - accuracy: 0.7107 - val_loss: 1.0892 - val_accuracy: 0.6465\n",
      "Epoch 17/35\n",
      "125/125 [==============================] - 43s 342ms/step - loss: 0.8240 - accuracy: 0.7179 - val_loss: 1.0541 - val_accuracy: 0.6510\n",
      "Epoch 18/35\n",
      "125/125 [==============================] - 43s 341ms/step - loss: 0.8014 - accuracy: 0.7281 - val_loss: 1.0807 - val_accuracy: 0.6565\n",
      "Epoch 19/35\n",
      "125/125 [==============================] - 42s 336ms/step - loss: 0.7858 - accuracy: 0.7278 - val_loss: 1.0469 - val_accuracy: 0.6590\n",
      "Epoch 20/35\n",
      "125/125 [==============================] - 43s 345ms/step - loss: 0.7583 - accuracy: 0.7398 - val_loss: 1.0351 - val_accuracy: 0.6540\n",
      "Epoch 21/35\n",
      "125/125 [==============================] - 43s 342ms/step - loss: 0.7457 - accuracy: 0.7451 - val_loss: 1.0618 - val_accuracy: 0.6505\n",
      "Epoch 22/35\n",
      "125/125 [==============================] - 43s 343ms/step - loss: 0.7269 - accuracy: 0.7550 - val_loss: 1.0743 - val_accuracy: 0.6565\n",
      "Epoch 23/35\n",
      "125/125 [==============================] - 43s 342ms/step - loss: 0.6977 - accuracy: 0.7644 - val_loss: 1.0671 - val_accuracy: 0.6535\n",
      "Epoch 24/35\n",
      "125/125 [==============================] - 42s 340ms/step - loss: 0.6770 - accuracy: 0.7707 - val_loss: 1.0756 - val_accuracy: 0.6610\n",
      "Epoch 25/35\n",
      "125/125 [==============================] - 43s 345ms/step - loss: 0.6602 - accuracy: 0.7793 - val_loss: 1.0874 - val_accuracy: 0.6500\n",
      "Epoch 26/35\n",
      "125/125 [==============================] - 42s 336ms/step - loss: 0.6578 - accuracy: 0.7775 - val_loss: 1.0579 - val_accuracy: 0.6525\n",
      "Epoch 27/35\n",
      "125/125 [==============================] - 43s 340ms/step - loss: 0.6265 - accuracy: 0.7886 - val_loss: 1.1005 - val_accuracy: 0.6580\n",
      "Epoch 28/35\n",
      "125/125 [==============================] - 43s 347ms/step - loss: 0.6070 - accuracy: 0.7950 - val_loss: 1.1193 - val_accuracy: 0.6625\n",
      "Epoch 29/35\n",
      "125/125 [==============================] - 43s 347ms/step - loss: 0.5874 - accuracy: 0.8043 - val_loss: 1.0826 - val_accuracy: 0.6625\n",
      "Epoch 30/35\n",
      "125/125 [==============================] - 43s 344ms/step - loss: 0.5704 - accuracy: 0.8052 - val_loss: 1.1182 - val_accuracy: 0.6585\n",
      "Epoch 31/35\n",
      "125/125 [==============================] - 43s 340ms/step - loss: 0.5439 - accuracy: 0.8161 - val_loss: 1.1162 - val_accuracy: 0.6695\n",
      "Epoch 32/35\n",
      "125/125 [==============================] - 41s 331ms/step - loss: 0.5388 - accuracy: 0.8170 - val_loss: 1.1287 - val_accuracy: 0.6665\n",
      "Epoch 33/35\n",
      "125/125 [==============================] - 42s 333ms/step - loss: 0.5191 - accuracy: 0.8215 - val_loss: 1.1426 - val_accuracy: 0.6665\n",
      "Epoch 34/35\n",
      "125/125 [==============================] - 41s 331ms/step - loss: 0.5013 - accuracy: 0.8307 - val_loss: 1.1590 - val_accuracy: 0.6630\n",
      "Epoch 35/35\n",
      "125/125 [==============================] - 41s 325ms/step - loss: 0.4786 - accuracy: 0.8375 - val_loss: 1.1569 - val_accuracy: 0.6630\n"
     ]
    }
   ],
   "source": [
    "# Train 3D CNN model-\n",
    "history = model.fit(\n",
    "    x = X_train, y = y_train,\n",
    "    batch_size = batch_size, epochs = 35,\n",
    "    validation_split = validation_split\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1bdb037a-4d81-402d-95dd-2b100bc4e794",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "73560b6a-aa54-4e52-8e7f-1220f01abe23",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save trained parameters-\n",
    "model.save_weights(\"Trained_3D_MNIST_model.h5\", overwrite = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ae53ad06-3f61-459d-81f2-a2b8f1e66154",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7f6f7fd3-326b-42c6-ab25-3c26d597be47",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "adcd70fe-f3d8-4d89-b25d-db0a0972d6f0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2eb18184-78ca-48dd-8853-7a375e0fb058",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "fbcef0e2-b24b-422b-a870-38725b0187f4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (8, 7))\n",
    "plt.plot(history.history['loss'], label = 'loss')\n",
    "plt.plot(history.history['val_loss'], label = 'val_loss')\n",
    "plt.xlabel(\"epochs\")\n",
    "plt.ylabel(\"loss\")\n",
    "plt.title(\"Training Visualization: Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "3caf6519-f42e-4bbc-9a35-c139205f5016",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (8, 7))\n",
    "plt.plot(history.history['accuracy'], label = 'acc')\n",
    "plt.plot(history.history['val_accuracy'], label = 'val_acc')\n",
    "plt.xlabel(\"epochs\")\n",
    "plt.ylabel(\"acc %\")\n",
    "plt.title(\"Training Visualization: Accuracy %\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b18eaf41-91a2-42b7-b5b1-c8d3d627d2d4",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "700fdb39-defd-4c0f-9641-6c5f381c69e0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "1e6d2a01-c5a6-4289-adec-198942f24c19",
   "metadata": {},
   "outputs": [],
   "source": [
    "import plotly.graph_objs as go\n",
    "# import plotly.plotly as py\n",
    "import plotly.express as px\n",
    "import plotly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "64840ea6-faae-47ca-8ba3-34beefb9f145",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "99d05299-2fe3-4c16-b3eb-b6a53a43aca1",
   "metadata": {},
   "outputs": [],
   "source": [
    "with h5py.File(\"3D_MNIST-archive/train_point_clouds.h5\", \"r\") as points_dataset:        \n",
    "    digits = []\n",
    "    for i in range(10):\n",
    "        digit = (\n",
    "            points_dataset[str(i)][\"img\"][:], \n",
    "            points_dataset[str(i)][\"points\"][:], \n",
    "            points_dataset[str(i)].attrs[\"label\"]\n",
    "        )\n",
    "        \n",
    "        digits.append(digit)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "20fc9d1f-7011-4792-bce3-ad58b0a08225",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(digits)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "b9f25d2a-a077-46d9-9f76-c1b1654f4a4b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(digits[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "305d120d-1d47-4ecd-b7d0-53b3080ede7d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((30, 30), (25700, 3), ())"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "digits[0][0].shape, digits[0][1].shape, digits[0][2].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "85080a30-25d8-406e-b4ac-abf87b35e7c4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30, 30)"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Img-\n",
    "digits[0][0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "81a6cd00-fd95-42cf-85b5-ef24827128db",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25700, 3)"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Points in point cloud-\n",
    "digits[0][1].shape\n",
    "# Consists of (x, y, z) coordinates, a tuple of length 3."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "b2f10751-0785-422c-b728-e02e7be2adc4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Label-\n",
    "digits[0][2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "c3a5c9b8-cede-4492-a823-075b151b180d",
   "metadata": {},
   "outputs": [],
   "source": [
    "for r in digits[0][1]:\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "c592db77-c7e1-4357-ac64-5a3fbeb7f26d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3,)"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "06b83ad3-1102-4fe0-8ce2-c12f5a3eeb5e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 0.15      , -0.1094358 ,  0.47305447])"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "efd899fd-86f6-41a9-80c4-78c324b4cb2e",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "1f5f8433-6644-44f3-86b6-d8667ed08ca6",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_c = [r[0] for r in digits[0][1]]\n",
    "y_c = [r[1] for r in digits[0][1]]\n",
    "z_c = [r[2] for r in digits[0][1]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "35cb8702-925b-4dd1-872a-33cc91d55aec",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25700, 25700, 25700)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(x_c), len(y_c), len(z_c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5a0e3e93-2cfb-447e-8eb2-0f4e46fa403d",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a75f66f2-484e-4c54-a658-a6401c7229b0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "08fa8077-92ac-429f-b7cb-0c415eed6637",
   "metadata": {},
   "outputs": [],
   "source": [
    "trace1 = go.Scatter3d(\n",
    "    x=x_c, y=y_c,\n",
    "    z=z_c, mode='markers', \n",
    "    marker = dict(\n",
    "        size=12, color=z_c,\n",
    "        colorscale='Viridis', opacity=0.7\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "14985275-8449-4797-8fa0-d1c7da16d8c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "data = [trace1]\n",
    "layout = go.Layout(\n",
    "    height = 500, width = 600,\n",
    "    title = \"Digit: \" + str(digits[0][2]) + \" in 3D space\"\n",
    ")\n",
    "fig = go.Figure(data = data, layout = layout)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b63f5e09-64fc-413f-a7d4-8ca475cee127",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c3600333-c617-44db-997d-765c8ea74e08",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e645c499-da02-4ca1-80bd-39ceab825d1c",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "45445870-a214-4fb5-b7c3-c57fb7e6cbd9",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "069e3d7a-d8ff-4eb0-889e-92a13fa2a2f5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25700, 3)"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "digits[0][1].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "cc62d6e4-8bf6-4c20-9526-e664eb4a77f1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(25700,)"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "digits[0][1][:, 0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7813c481-4b4f-4d02-b873-09e631059bbe",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ca88fb8a-0c58-46b3-b097-43fa3d528759",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = px.scatter_3d(\n",
    "    digits[0][1], x = digits[0][1][:, 0],\n",
    "    y = digits[0][1][:, 1], z = digits[0][1][:, 2],\n",
    "    color = digits[0][1][:, 2]\n",
    ")\n",
    "\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9ce114a5-decd-4aef-86d6-cfd4d7029b61",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "15f7e76f-a194-47df-b8c8-dad37a4525ca",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}