neural network for high school

Author: Satyam19946

WI21/nn.ipynb

@@ -0,0 +1,186 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 50,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from keras.models import Sequential\n",
+    "from keras.layers import Dense, Flatten\n",
+    "from keras.layers import BatchNormalization as BatchNorm\n",
+    "from keras.utils import np_utils\n",
+    "from keras.callbacks import ModelCheckpoint\n",
+    "import random\n",
+    "import tensorflow as tf"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 48,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(60000, 28, 28)\n"
+     ]
+    }
+   ],
+   "source": [
+    "(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data(path=\"mnist.npz\")\n",
+    "print(x_train.shape)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 49,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(42000, 28, 28) (18000, 28, 28)\n"
+     ]
+    }
+   ],
+   "source": [
+    "# train, validation split (70%, 30%)\n",
+    "x_train, x_val = x_train[0:x_train.shape[0]*70//100], x_train[x_train.shape[0]*70//100:]\n",
+    "y_train, y_val = y_train[0:y_train.shape[0]*70//100], y_train[y_train.shape[0]*70//100:]\n",
+    "# Some other simple things you can do - Shuffle, Normalize. \n",
+    "# Other advanced things you can do to preprocess the data - PCA, Z-score.\n",
+    "# What other optimizations can you think of? \n",
+    "print(x_train.shape, x_val.shape)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 58,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# https://keras.io/api/layers/\n",
+    "# Keras offers an API for a lot of layers with multiple optional parameters to tune the network.\n",
+    "def create_network(network_input):\n",
+    "    model = Sequential()\n",
+    "    model.add(Flatten()) # Convert [28,28] -> [784,]\n",
+    "    model.add(Dense(25)) # [784,] -> [25,]\n",
+    "    model.add(Activation('relu'))\n",
+    "    model.add(Dense(10)) # [25,] -> [10,] FCC\n",
+    "    model.add(Activation('softmax'))\n",
+    "    \n",
+    "    #optimizer and loss.\n",
+    "    model.compile(optimizer='adam', loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True))\n",
+    "    return model\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 60,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/5\n",
+      "5250/5250 [==============================] - 3s 538us/step - loss: 3.6693\n",
+      "Epoch 2/5\n",
+      "5250/5250 [==============================] - 3s 534us/step - loss: 0.9733\n",
+      "Epoch 3/5\n",
+      "5250/5250 [==============================] - 3s 543us/step - loss: 0.7850\n",
+      "Epoch 4/5\n",
+      "5250/5250 [==============================] - 3s 537us/step - loss: 0.6917\n",
+      "Epoch 5/5\n",
+      "5250/5250 [==============================] - 3s 544us/step - loss: 0.6751\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<tensorflow.python.keras.callbacks.History at 0x7fcd40099fd0>"
+      ]
+     },
+     "execution_count": 60,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "model = create_network(x_train)\n",
+    "#https://keras.io/api/models/model_training_apis/\n",
+    "#without validation\n",
+    "model.fit(x=x_train, y=y_train, epochs=5, batch_size = 8)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 61,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Epoch 1/5\n",
+      "5250/5250 [==============================] - 4s 682us/step - loss: 4.4919 - val_loss: 1.3123\n",
+      "Epoch 2/5\n",
+      "5250/5250 [==============================] - 4s 669us/step - loss: 1.2269 - val_loss: 0.9553\n",
+      "Epoch 3/5\n",
+      "5250/5250 [==============================] - 4s 676us/step - loss: 0.9047 - val_loss: 0.7206\n",
+      "Epoch 4/5\n",
+      "5250/5250 [==============================] - 3s 666us/step - loss: 0.7218 - val_loss: 0.7201\n",
+      "Epoch 5/5\n",
+      "5250/5250 [==============================] - 4s 673us/step - loss: 0.6267 - val_loss: 0.6180\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<tensorflow.python.keras.callbacks.History at 0x7fcd307d9070>"
+      ]
+     },
+     "execution_count": 61,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "model = create_network(x_train)\n",
+    "#with validation\n",
+    "model.fit(x=x_train, y=y_train, epochs=5, batch_size = 8, validation_data=(x_val, y_val))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "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": "3.8.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}