week-two.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# __Week 2__: Neural Networks\n",
    "***"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Biological Motivation\n",
    "- The basic computational unit of the brain is a __neuron__, and they are connected by __synapses__. Each neuron recieves input signals from its __dendrites__ and produces output signals along its __axon__.\n",
    "\n",
    "<img src=\"figures/neuron.png\" alt=\"Neuron ex\" title=\"Bio Motivation\" width=\"500\"/>\n",
    "\n",
    "- Neural networks with at least one hidden layer are <i>universal approximators</i>. This simply means that given any continuous function $f(x)$ and some $\\epsilon > 0$, there exists a Neural Network $g(x)$ with one hidden layer such that $\\forall, |f(x) - g(x)| < \\epsilon$. In other words, __the Neural Network can approximate any continuous function__.\n",
    "\n",
    "- Neural Networks work well in practice because they compactly express nice, smooth functions that fit well with the statistical properties of data we encounter in practice.\n",
    "\n",
    "- Larger Neural Networks can represent more complicated functions.\n",
    "\n",
    "<img src=\"figures/layers.jpg\" alt=\"Complicated\" title=\"Hidden neurons\" width=\"500\"/>\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Preprocessing\n",
    "- *Previous lecture\n",
    "- Dimension Reduction\n",
    "    - Principal Component Analysis\n",
    "- Normalization - normalizing the data dimensions so that they are of approximately the same scale"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Neural Network Architectures\n",
    "- Fully- vs. Partially- Connected layers\n",
    "    - Fully - neurons between two adjacent layers are fully pairwise connected, but neurons within a single layer share no connections\n",
    "- Naming Conventions:\n",
    "    - N-layer NN -- we don't count the input layer\n",
    "        - Also called ANNs (Artificial Neural Networks) or MLPs (Multilayer Perceptrons)\n",
    "- Output layer - output layer neurons most commonly do not have an activation function\n",
    "    - Usually taken to represent class scores or real-valued target\n",
    "- The forward pass of a fully-connected layer corresponds to one matrix multiplication followed by a bias offset and an activation function.\n",
    "\n",
    "<img src=\"figures/nn.jpg\" alt=\"NN\" title=\"neurons\" width=\"250\"/>\n",
    "<img src=\"figures/nn2.jpg\" alt=\"NN\" title=\"neurons\" width=\"250\"/>\n",
    "\n",
    "\n",
    "#### Activation Functions - takes a single number and performs a certain fixed mathematical operation on it. \n",
    "- Sigmoid, Tanh, ReLU, Leaky ReLU, Maxout, etc.\n",
    "- Are placed <i>between</i> layers.\n",
    "\n",
    "\n",
    "#### Other relevant vocab\n",
    "- Overfitting - occurs when a model with high capacity fits the <i>noise</i> in the data instead of the assumed underlying relationship\n",
    "- Epoch - number of passes of entire training dataset through ML algorithm \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loss Functions\n",
    "- Data loss - measures quality of a particular set of parameters based on how well induced scores agree with ground truth labels\n",
    "- Mean squared error, Hinge loss, Cross Entropy Loss, etc.\n",
    "- Should decrease as epochs advance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learning Process/Parameter Updates\n",
    "- Optimization - process of finding the set of parameters $W$ that minimize the loss function\n",
    "    - Want to advance in direction that improves parameters (this is mathematically guaranteed to be the direction of steepest descent), which is related to the __gradient__ of the loss function"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Motivating Example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import sys\n",
    "import re\n",
    "from nltk.tokenize import RegexpTokenizer\n",
    "from nltk.corpus import stopwords"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Poe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "file = open(\"../poe_allen.txt\").read()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "a = file.split('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "for item in a:\n",
    "    if len(item) <= 5:\n",
    "        a.remove(item)\n",
    "    else:\n",
    "        continue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "f_clean = []\n",
    "\n",
    "for episode in a:\n",
    "    f_clean.append(re.sub(r'[^\\w]', ' ', episode))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "texts = f_clean.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Shakespeare"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "file2 = open(\"../shakespeare.txt\").read()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "a2 = file2.split('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "for item in a2:\n",
    "    if len(item) <= 5:\n",
    "        a2.remove(item)\n",
    "    else:\n",
    "        continue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "f_clean2 = []\n",
    "\n",
    "for episode in a2:\n",
    "    f_clean2.append(re.sub(r'[^\\w]', ' ', episode))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "texts2 = f_clean2.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Creating Sentence Embeddings using BERT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sentence_transformers import SentenceTransformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3f177ad3801341ec9bc60b775847b945",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=391.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5144836bdbc9477581c52caff6660752",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=3931.0, style=ProgressStyle(description…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ba806d5d34ae403cb3eea1eada8aacb0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=2.0, style=ProgressStyle(description_wi…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "381be16306974f3bb69904599887b02b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=625.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d67672b7fd574574ab17fa2aeca83f83",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=122.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e0fc4344b1bf4dbabc01c938b2a26196",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=229.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fe6a16b5e4eb4c81bf076cade1bd0728",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=438007537.0, style=ProgressStyle(descri…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "438dabaf07634c919bf13e2acd65a9f6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=53.0, style=ProgressStyle(description_w…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f7a4045a93974a8a81000d12c0319b49",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=112.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "282795497bd54814819608f7fb604e8a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=466081.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f2619873e6874abf8f27b77d444a294f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=399.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9c96692f1495482284f7aa038eb9c5ab",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=231508.0, style=ProgressStyle(descripti…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "615d51622a0744a99e375ed5e0c3f859",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, description='Downloading', max=190.0, style=ProgressStyle(description_…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "[[-0.39310005  0.03886304  1.9874251  ... -0.6093677  -1.094621\n",
      "   0.3264901 ]\n",
      " [ 0.06153379  0.32736215  1.8332328  ... -0.12985346  0.46089378\n",
      "   0.24035406]]\n"
     ]
    }
   ],
   "source": [
    "model = SentenceTransformer('sentence-transformers/bert-base-nli-mean-tokens')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "embeddings = model.encode(texts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8919, 768)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embeddings.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "embeddings2 = model.encode(texts2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2236, 768)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embeddings2.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Visualizing Embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.manifold import TSNE\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "N=250"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "number_of_rows = embeddings.shape[0]\n",
    "random_indices = np.random.choice(number_of_rows, size=N, replace=False)\n",
    "random_rows = embeddings[random_indices, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_embedded = TSNE(n_components=2,random_state=0).fit_transform(random_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "number_of_rows2 = embeddings2.shape[0]\n",
    "random_indices2 = np.random.choice(number_of_rows2, size=N, replace=False)\n",
    "random_rows2 = embeddings2[random_indices2, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_embedded2 = TSNE(n_components=2,random_state=0).fit_transform(random_rows2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_subset = pd.DataFrame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_subset['tsne-2d-one'] = X_embedded[:,0]\n",
    "df_subset['tsne-2d-two'] = X_embedded[:,1]\n",
    "plt.figure(figsize=(16,10))\n",
    "sns.scatterplot(\n",
    "    x=\"tsne-2d-one\", y=\"tsne-2d-two\",\n",
    "    data=df_subset,\n",
    "    color=\"green\",\n",
    "    legend=\"full\",\n",
    "    alpha=0.5\n",
    ")\n",
    "\n",
    "df_subset['tsne-2d-one2'] = X_embedded2[:,0]\n",
    "df_subset['tsne-2d-two2'] = X_embedded2[:,1]\n",
    "sns.scatterplot(\n",
    "    x=\"tsne-2d-one2\", y=\"tsne-2d-two2\",\n",
    "    data=df_subset,\n",
    "    color=\"magenta\",\n",
    "    legend=\"full\",\n",
    "    alpha=0.3\n",
    ")\n",
    "\n",
    "plt.legend(['Edgar Allen Poe', 'William Shakespeare'], fontsize=18)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sources\n",
    "[1] http://web.stanford.edu/class/cs224n/index.html#schedule <br>\n",
    "[2] https://cs231n.github.io <br>\n",
    "[3] https://www.kaggle.com/mruanova/shakespeare <br>\n",
    "[4] https://www.kaggle.com/tiaschmidt/edgar-allan-poes-complete-work <br>\n",
    "[5] https://huggingface.co/sentence-transformers/bert-base-nli-mean-tokens <br>\n",
    "[6] https://towardsdatascience.com/visualising-high-dimensional-datasets-using-pca-and-t-sne-in-python-8ef87e7915b <br>"
   ]
  },
  {
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}