example_basic_usage.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# HyperbolicTSNE\n",
    "\n",
    "This notebook illustrates the usage of the HyperbolicTSNE library. Specifically, we load a subset of the MNIST dataset and embed it in hyperbolic space using the accelerated version of hyperbolic tsne. Finally, we save the embedding result as an image."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup\n",
    "\n",
    "First, we import the packages we will use and set important paths. Note that `hyperbolicTSNE.util` and `hyperbolicTSNE.visualization` contain useful functions for reading, processing and exporting embeddings. This requires that hyperbolicTSNE has been set up as detailed in the main readme of the repository. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:18:09.224771636Z",
     "start_time": "2023-11-13T15:18:08.516888492Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Please note that `empty_sequence` uses the KL divergence with Barnes-Hut approximation (angle=0.5) by default.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import traceback\n",
    "\n",
    "from hyperbolicTSNE.util import find_last_embedding\n",
    "from hyperbolicTSNE.visualization import plot_poincare, animate\n",
    "from hyperbolicTSNE import load_data, Datasets, SequentialOptimizer, initialization, HyperbolicTSNE"
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "We assume that there is a top-level folder `datasets` that holds the MNIST data set. Refer to the main readme of the repository for where to find the data sets used in this repository."
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:20:09.564819212Z",
     "start_time": "2023-11-13T15:20:09.560922527Z"
    }
   },
   "outputs": [],
   "source": [
    "data_home = \"datasets\"\n",
    "log_path = \"temp/poincare/\"  # path for saving embedding snapshots"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Configure\n",
    "\n",
    "HyperbolicTSNE follows a similar API to other t-SNE libraries like OpenTSNE and sklearn. The configuration process consists of loading the data to embed and defining the settings of the embedder. We create a dict with parameters manually to demonstrate all the customization options. Nevertheless, `hyperbolicTSNE.hyperbolicTSNE` provides parameter templates to start with."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:25:10.788279509Z",
     "start_time": "2023-11-13T15:25:07.900874773Z"
    }
   },
   "outputs": [],
   "source": [
    "only_animate = False\n",
    "seed = 42\n",
    "dataset = Datasets.MNIST  # the Datasets handler provides access to several data sets used throughout the repository\n",
    "num_points = 10000  # we use a subset for demonstration purposes, full MNIST has N=70000\n",
    "perp = 30  # we use a perplexity of 30 in this example\n",
    "\n",
    "dataX, dataLabels, D, V, _ = load_data(\n",
    "    dataset, \n",
    "    data_home=data_home, \n",
    "    random_state=seed, \n",
    "    to_return=\"X_labels_D_V\",\n",
    "    hd_params={\"perplexity\": perp}, \n",
    "    sample=num_points, \n",
    "    knn_method=\"hnswlib\"  # we use an approximation of high-dimensional neighbors to speed up computations\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:37:10.158602643Z",
     "start_time": "2023-11-13T15:37:10.142279339Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Please note that `empty_sequence` uses the KL divergence with Barnes-Hut approximation (angle=0.5) by default.\n",
      "config: {'learning_rate_ex': 0.8333333333333334, 'learning_rate_main': 0.8333333333333334, 'exaggeration': 12, 'exaggeration_its': 250, 'gradientDescent_its': 750, 'vanilla': False, 'momentum_ex': 0.5, 'momentum': 0.8, 'exact': False, 'area_split': False, 'n_iter_check': 10, 'size_tol': 0.999}\n"
     ]
    }
   ],
   "source": [
    "exaggeration_factor = 12  # Just like regular t-SNE, we use early exaggeration with a factor of 12\n",
    "learning_rate = (dataX.shape[0] * 1) / (exaggeration_factor * 1000)  # We adjust the learning rate to the hyperbolic setting\n",
    "ex_iterations = 250  # The embedder is to execute 250 iterations of early exaggeration, ...\n",
    "main_iterations = 750  # ... followed by 750 iterations of non-exaggerated gradient descent.\n",
    "\n",
    "opt_config = dict(\n",
    "    learning_rate_ex=learning_rate,  # learning rate during exaggeration\n",
    "    learning_rate_main=learning_rate,  # learning rate main optimization \n",
    "    exaggeration=exaggeration_factor, \n",
    "    exaggeration_its=ex_iterations, \n",
    "    gradientDescent_its=main_iterations, \n",
    "    vanilla=False,  # if vanilla is set to true, regular gradient descent without any modifications is performed; for  vanilla set to false, the optimization makes use of momentum and gains\n",
    "    momentum_ex=0.5,  # Set momentum during early exaggeration to 0.5\n",
    "    momentum=0.8,  # Set momentum during non-exaggerated gradient descent to 0.8\n",
    "    exact=False,  # To use the quad tree for acceleration (like Barnes-Hut in the Euclidean setting) or to evaluate the gradient exactly\n",
    "    area_split=False,  # To build or not build the polar quad tree based on equal area splitting or - alternatively - on equal length splitting\n",
    "    n_iter_check=10,  # Needed for early stopping criterion\n",
    "    size_tol=0.999  # Size of the embedding to be used as early stopping criterion\n",
    ")\n",
    "\n",
    "opt_params = SequentialOptimizer.sequence_poincare(**opt_config)\n",
    "\n",
    "# Start: configure logging\n",
    "logging_dict = {\n",
    "    \"log_path\": log_path\n",
    "}\n",
    "opt_params[\"logging_dict\"] = logging_dict\n",
    "\n",
    "log_path = opt_params[\"logging_dict\"][\"log_path\"]\n",
    "# Delete old log path\n",
    "if os.path.exists(log_path) and not only_animate:\n",
    "    import shutil\n",
    "    shutil.rmtree(log_path)\n",
    "# End: logging\n",
    "\n",
    "print(f\"config: {opt_config}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run HyperbolicTSNE\n",
    "\n",
    "Embedding the high dimensional data consists of three steps:\n",
    "- Initializating the embedding\n",
    "- Initializing the embedder \n",
    "- Embedding the data\n",
    "\n",
    "The following three cells demonstrate this process. Note that use set metric to \"precomputed\" because we pass the distance matrix to the `fit` method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:37:39.820114839Z",
     "start_time": "2023-11-13T15:37:39.690680821Z"
    }
   },
   "outputs": [],
   "source": [
    "# Compute an initial embedding of the data via PCA\n",
    "X_embedded = initialization(\n",
    "    n_samples=dataX.shape[0],\n",
    "    n_components=2,\n",
    "    X=dataX,\n",
    "    random_state=seed,\n",
    "    method=\"pca\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:38:18.116323976Z",
     "start_time": "2023-11-13T15:38:18.075397174Z"
    }
   },
   "outputs": [],
   "source": [
    "# Initialize the embedder\n",
    "htsne = HyperbolicTSNE(\n",
    "    init=X_embedded, \n",
    "    n_components=2, \n",
    "    metric=\"precomputed\", \n",
    "    verbose=True, \n",
    "    opt_method=SequentialOptimizer, \n",
    "    opt_params=opt_params\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:40:06.268727695Z",
     "start_time": "2023-11-13T15:38:58.361025102Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[HyperbolicTSNE] Received iterable as input. It should have len=2 and contain (D=None, V=None)\n",
      "[hd_mat] Warning: There is nothing to do with given parameters. Returning given D and V\n",
      "Running Gradient Descent, Verbosity: True\n",
      "[gradient_descent] Warning: because of logging, the cf will be computed at every iteration\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Gradient Descent error: 97.24348 grad_norm: 6.52115e-01: 100%|██████████| 250/250 [00:30<00:00,  8.25it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running Gradient Descent, Verbosity: True\n",
      "[gradient_descent] Warning: because of logging, the cf will be computed at every iteration\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Gradient Descent error: 4.04533 grad_norm: 1.99173e+05:  39%|███▊      | 289/750 [00:37<00:59,  7.69it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Compute the embedding\n",
    "try:\n",
    "    hyperbolicEmbedding = htsne.fit_transform((D, V))\n",
    "except ValueError:\n",
    "    hyperbolicEmbedding = find_last_embedding(log_path)\n",
    "    traceback.print_exc()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Exporting and visualization\n",
    "\n",
    "After running the embedding process, the embeddings arrays are saved to the `log_path`. We can use this information to visualize the embeddings using utility functions defined in `hyperbolicTSNE.visualization` as shown below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:43:29.644001757Z",
     "start_time": "2023-11-13T15:43:29.408874430Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create a rendering of the embedding and save it to a file\n",
    "if not os.path.exists(\"results\"):\n",
    "    os.mkdir(\"results\")\n",
    "fig = plot_poincare(hyperbolicEmbedding, dataLabels)\n",
    "fig.savefig(f\"results/{dataset.name}.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-11-13T15:49:11.530204393Z",
     "start_time": "2023-11-13T15:49:04.075561907Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Animation being saved to: results/MNIST_ani.mp4\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "Animating:   0%|          | 0/54 [00:00<?, ?it/s]\u001B[A/home/martin/Projects/hyperbolic-tsne/hyperbolicTSNE/visualization.py:318: UserWarning: You passed in an explicit save_count=50 which is being ignored in favor of frames=54.\n",
      "  anim = FuncAnimation(fig, update, frames=len(scatter_data), interval=50, blit=True, save_count=50)\n",
      "\n",
      "Animating:   9%|▉         | 5/54 [00:00<00:02, 19.41it/s]\u001B[A\n",
      "Animating:  13%|█▎        | 7/54 [00:00<00:03, 12.95it/s]\u001B[A\n",
      "Animating:  17%|█▋        | 9/54 [00:00<00:04, 11.16it/s]\u001B[A\n",
      "Animating:  20%|██        | 11/54 [00:00<00:04,  9.96it/s]\u001B[A\n",
      "Animating:  22%|██▏       | 12/54 [00:01<00:04,  9.68it/s]\u001B[A\n",
      "Animating:  24%|██▍       | 13/54 [00:01<00:04,  9.38it/s]\u001B[A\n",
      "Animating:  26%|██▌       | 14/54 [00:01<00:04,  9.22it/s]\u001B[A\n",
      "Animating:  28%|██▊       | 15/54 [00:01<00:04,  9.03it/s]\u001B[A\n",
      "Animating:  30%|██▉       | 16/54 [00:01<00:04,  8.73it/s]\u001B[A\n",
      "Animating:  31%|███▏      | 17/54 [00:01<00:04,  8.64it/s]\u001B[A\n",
      "Animating:  33%|███▎      | 18/54 [00:01<00:04,  8.61it/s]\u001B[A\n",
      "Animating:  35%|███▌      | 19/54 [00:01<00:04,  8.62it/s]\u001B[A\n",
      "Animating:  37%|███▋      | 20/54 [00:02<00:03,  8.59it/s]\u001B[A\n",
      "Animating:  39%|███▉      | 21/54 [00:02<00:03,  8.49it/s]\u001B[A\n",
      "Animating:  41%|████      | 22/54 [00:02<00:03,  8.44it/s]\u001B[A\n",
      "Animating:  43%|████▎     | 23/54 [00:02<00:03,  8.46it/s]\u001B[A\n",
      "Animating:  44%|████▍     | 24/54 [00:02<00:03,  8.52it/s]\u001B[A\n",
      "Animating:  46%|████▋     | 25/54 [00:02<00:03,  8.48it/s]\u001B[A\n",
      "Animating:  48%|████▊     | 26/54 [00:02<00:03,  8.27it/s]\u001B[A\n",
      "Animating:  50%|█████     | 27/54 [00:02<00:03,  7.27it/s]\u001B[A\n",
      "Animating:  52%|█████▏    | 28/54 [00:03<00:03,  7.65it/s]\u001B[A\n",
      "Animating:  54%|█████▎    | 29/54 [00:03<00:03,  7.89it/s]\u001B[A\n",
      "Animating:  56%|█████▌    | 30/54 [00:03<00:03,  7.74it/s]\u001B[A\n",
      "Animating:  57%|█████▋    | 31/54 [00:03<00:02,  8.04it/s]\u001B[A\n",
      "Animating:  59%|█████▉    | 32/54 [00:03<00:02,  8.07it/s]\u001B[A\n",
      "Animating:  61%|██████    | 33/54 [00:03<00:02,  8.16it/s]\u001B[A\n",
      "Animating:  63%|██████▎   | 34/54 [00:03<00:02,  8.27it/s]\u001B[A\n",
      "Animating:  65%|██████▍   | 35/54 [00:03<00:02,  8.35it/s]\u001B[A\n",
      "Animating:  67%|██████▋   | 36/54 [00:04<00:02,  8.34it/s]\u001B[A\n",
      "Animating:  69%|██████▊   | 37/54 [00:04<00:02,  8.18it/s]\u001B[A\n",
      "Animating:  70%|███████   | 38/54 [00:04<00:01,  8.28it/s]\u001B[A\n",
      "Animating:  72%|███████▏  | 39/54 [00:04<00:01,  8.11it/s]\u001B[A\n",
      "Animating:  74%|███████▍  | 40/54 [00:04<00:01,  8.26it/s]\u001B[A\n",
      "Animating:  76%|███████▌  | 41/54 [00:04<00:01,  8.29it/s]\u001B[A\n",
      "Animating:  78%|███████▊  | 42/54 [00:04<00:01,  8.28it/s]\u001B[A\n",
      "Animating:  80%|███████▉  | 43/54 [00:04<00:01,  8.19it/s]\u001B[A\n",
      "Animating:  81%|████████▏ | 44/54 [00:04<00:01,  8.30it/s]\u001B[A\n",
      "Animating:  83%|████████▎ | 45/54 [00:05<00:01,  8.42it/s]\u001B[A\n",
      "Animating:  85%|████████▌ | 46/54 [00:05<00:00,  8.48it/s]\u001B[A\n",
      "Animating:  87%|████████▋ | 47/54 [00:05<00:00,  8.43it/s]\u001B[A\n",
      "Animating:  89%|████████▉ | 48/54 [00:05<00:00,  8.48it/s]\u001B[A\n",
      "Animating:  91%|█████████ | 49/54 [00:05<00:00,  8.55it/s]\u001B[A\n",
      "Animating:  93%|█████████▎| 50/54 [00:05<00:00,  8.58it/s]\u001B[A\n",
      "Animating:  94%|█████████▍| 51/54 [00:05<00:00,  8.14it/s]\u001B[A\n",
      "Animating:  96%|█████████▋| 52/54 [00:05<00:00,  8.16it/s]\u001B[A\n",
      "Animating:  98%|█████████▊| 53/54 [00:06<00:00,  8.21it/s]\u001B[A\n",
      "Animating: 100%|██████████| 54/54 [00:06<00:00,  7.95it/s]\u001B[A\n",
      "Animating: : 55it [00:06,  8.11it/s]                      \u001B[A\n",
      "Animating: : 56it [00:06,  8.20it/s]\u001B[A\n",
      "Animating: : 57it [00:06,  8.45it/s]\u001B[A\n"
     ]
    }
   ],
   "source": [
    "# This renders a GIF animation of the embedding process. If FFMPEG is installed, the command also supports .mp4 as file ending \n",
    "animate(logging_dict, dataLabels, f\"results/{dataset.name}_ani.gif\", fast=True, plot_ee=True)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "htsne",
   "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": "3.9.16"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}