rank_coding_mnist.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "Rax5p4AtkZGW"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.lines import Line2D\n",
    "from matplotlib.patches import Patch\n",
    "\n",
    "import torch.nn.functional as F\n",
    "from torch import nn\n",
    "from torchvision import datasets, transforms\n",
    "from tqdm.notebook import tqdm\n",
    "from tqdm import tqdm\n",
    "import time\n",
    "\n",
    "device = 'cuda' if torch.cuda.is_available() else 'cpu'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "mOaV76G6kb1o"
   },
   "outputs": [],
   "source": [
    "def data_generator(root, batch_size):\n",
    "    train_set = datasets.MNIST(root=root, train=True, download=True,\n",
    "                               transform=transforms.Compose([\n",
    "                                   transforms.ToTensor()\n",
    "                                  #  transforms.Normalize((0.1307,), (0.3081,))\n",
    "                               ]))\n",
    "    test_set = datasets.MNIST(root=root, train=False, download=True,\n",
    "                              transform=transforms.Compose([\n",
    "                                  transforms.ToTensor()\n",
    "                                  # transforms.Normalize((0.1307,), (0.3081,))\n",
    "                              ]))\n",
    "\n",
    "    train_loader = torch.utils.data.DataLoader(train_set, batch_size=batch_size)\n",
    "    test_loader = torch.utils.data.DataLoader(test_set, batch_size=batch_size)\n",
    "    return train_loader, test_loader\n",
    "\n",
    "# download data\n",
    "train_loader, test_loader = data_generator('/', 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "wEjHP6J8kede"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "_NujaHrqWs0F"
   },
   "outputs": [],
   "source": [
    "def to_temporal(batch, step=0.1):\n",
    "  \"\"\"\n",
    "  Convert a batch of mnist images into temporal format with timing\n",
    "  determined by the intensity of the pixel from darkest to lightest.\n",
    "\n",
    "  Args:\n",
    "      batch (torch tensor): batch of mnist images.\n",
    "      step (float): determines the number of time intervals with number\n",
    "          of intervals given by 1/step.\n",
    "  \n",
    "  Returns:\n",
    "      torch tensor of the batch of images in temporal format.\n",
    "  \"\"\"\n",
    "  buckets, _ = torch.sort(1 - torch.arange(start=0, end=1, step=step))\n",
    "  # bucketize into desired number of temporal bins\n",
    "  bucket_batch = torch.bucketize(batch, buckets)\n",
    "  # one hot encode these bins and flip so that dark appears first\n",
    "  temp_batch = torch.flip(F.one_hot(bucket_batch), [4])\n",
    "  return temp_batch\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "j0aErTq-yhrI",
    "outputId": "7255d61b-6a77-455f-b8a9-830e9398a69f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original Image\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Temporal version of image\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot a sample temporal image\n",
    "for batch_idx, (data, target) in enumerate(train_loader):\n",
    "  break\n",
    "print('Original Image')\n",
    "plt.imshow(data[0].squeeze(0).numpy())\n",
    "plt.title('Original Image')\n",
    "plt.show()\n",
    "test = to_temporal(data)\n",
    "print('\\nTemporal version of image')\n",
    "for i in range(10):\n",
    "  x = test[0,0,:,:,i].numpy()\n",
    "  plt.imshow(x)\n",
    "  plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "9ETK4xKDi5aG"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "9nbpnztCcTja"
   },
   "outputs": [],
   "source": [
    "# LSTM model\n",
    "torch.manual_seed(3)\n",
    "\n",
    "class LSTM(nn.Module):\n",
    "    def __init__(self, input_size, output_size, nhid, f_bias=True):\n",
    "        super(LSTM, self).__init__()\n",
    "        self.nhid = nhid\n",
    "        self.lstm = nn.LSTM(input_size, nhid, batch_first=True)\n",
    "        self.linear = nn.Linear(nhid, output_size)\n",
    "\n",
    "    def forward(self, inputs):\n",
    "        \"\"\"Inputs have to have dimension (N, C_in, L_in)\"\"\"\n",
    "        h0 = torch.zeros(1, inputs.size(0), self.nhid).to(device) \n",
    "        c0 = torch.zeros(1, inputs.size(0), self.nhid).to(device)\n",
    "        y1, _ = self.lstm(inputs, (h0, c0))  # input should have dimension (B, S, I)\n",
    "        o = self.linear(y1)\n",
    "        return o"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "7Z7BM-s-kerz",
    "outputId": "8e59d075-6d1c-491d-9f74-6617a7b2d52f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "################## TRAINING ##################\n",
      "Training 30 epochs in batches of size 256\n",
      "##############################################\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 30/30 [35:56<00:00, 71.89s/it]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Final loss: 0.95419921875\n"
     ]
    }
   ],
   "source": [
    "model = LSTM(28*28, 10, 125)\n",
    "model.to(device)\n",
    "optimiser = torch.optim.Adam(model.parameters(), lr=0.001)\n",
    "criterion = torch.nn.CrossEntropyLoss()\n",
    "s = nn.Sigmoid()\n",
    "\n",
    "BATCH_SIZE = 256\n",
    "VALIDATION_FREQ = 1\n",
    "PATH = '/model_weights'\n",
    "\n",
    "NUM_EPOCHS = 30\n",
    "seq_len = 20\n",
    "train_loader, test_loader = data_generator('/', BATCH_SIZE)\n",
    "\n",
    "best_spiking_acc = 0\n",
    "loss_ls = []\n",
    "spiking_acc_ls = []\n",
    "spike_time_ls = []\n",
    "\n",
    "print('################## TRAINING ##################')\n",
    "print('Training {} epochs in batches of size {}'.format(NUM_EPOCHS, BATCH_SIZE))\n",
    "print('##############################################\\n')\n",
    "\n",
    "time.sleep(0.2)\n",
    "for epoch in tqdm(range(NUM_EPOCHS)):\n",
    "    for batch_idx, (batch, targets) in enumerate(train_loader):\n",
    "        ######################## TRAIN A BATCH ########################\n",
    "        batch_size = batch.size(0)\n",
    "        temp_batch = to_temporal(batch, step=0.05)\n",
    "        temp_batch = temp_batch.view(batch_size, 28*28, seq_len).permute(0,2,1)\n",
    "        temp_batch, targets = temp_batch.to(device), targets.to(device)\n",
    "\n",
    "        # learn through network\n",
    "        model.train()\n",
    "        optimiser.zero_grad()\n",
    "        out = model(temp_batch.float())\n",
    "        out = s(out)\n",
    "        \n",
    "        # grab indixes of first spikes\n",
    "        with torch.no_grad():\n",
    "            spikes = (out > 0.95) * 1\n",
    "            spikes_red = spikes.sum(axis=2)  # find rows with no spikes\n",
    "            # set seqs with no spikes to spike on last output\n",
    "            spikes_red[torch.sum(spikes_red, axis=1) == 0, -1] = 1\n",
    "            # find indexes of first spike in seqs and pass it back\n",
    "            first_spike_idx = torch.argmax(spikes_red, dim=1, keepdim=True)\n",
    "            first_spike_idx = first_spike_idx.repeat(1, 10).unsqueeze(1)\n",
    " \n",
    "        preds = torch.gather(out, 1, first_spike_idx)  \n",
    "        preds = preds.squeeze(1)\n",
    "        \n",
    "        loss = criterion(preds, targets)\n",
    "        loss_ls.append(loss.item())\n",
    "        loss.backward()\n",
    "        optimiser.step()\n",
    "\n",
    "    ######################## VALIDATION SET ########################\n",
    "    if epoch % VALIDATION_FREQ == 0:\n",
    "      temp_spiking_acc_ls = []\n",
    "      temp_spike_time_ls = []\n",
    "      model.eval()\n",
    "      with torch.no_grad():\n",
    "          for batch, targets in test_loader:\n",
    "              batch_size = batch.size(0)\n",
    "              temp_batch = to_temporal(batch, step=0.05)\n",
    "              temp_batch = temp_batch.view(batch_size, 28*28, seq_len).permute(0,2,1)\n",
    "              temp_batch, targets = temp_batch.to(device), targets.to(device)\n",
    "\n",
    "              out = model(temp_batch.float())\n",
    "              out = s(out)\n",
    "              \n",
    "              # generate tensor of spikes\n",
    "              spikes = (out > 0.95) * 1\n",
    "\n",
    "              spikes_red = spikes.sum(axis=2)  # find rows with no spikes\n",
    "              # set seqs with no spikes to spike on last output\n",
    "              spikes_red[torch.sum(spikes_red, axis=1) == 0, -1] = 1\n",
    "              # find indexes of first spike in seqs and pass it back\n",
    "              first_spike_idx = torch.argmax(spikes_red, dim=1, keepdim=True)\n",
    "              batch_spike_time = torch.mean(first_spike_idx.float())\n",
    "              first_spike_idx = first_spike_idx.repeat(1, 10).unsqueeze(1)\n",
    "  \n",
    "              preds = torch.gather(out, 1, first_spike_idx)\n",
    "              preds = preds.squeeze(1)\n",
    "\n",
    "              preds = torch.argmax(preds, dim=1)\n",
    "\n",
    "              # spiking accuracy\n",
    "              spiking_acc_val = (torch.sum(targets == preds)/batch_size).item()\n",
    "              temp_spiking_acc_ls.append(spiking_acc_val)\n",
    "\n",
    "              # store mean spike times\n",
    "              temp_spike_time_ls.append(batch_spike_time.item())\n",
    "\n",
    "      # save model if better than current best\n",
    "      val_set_spiking_acc = np.mean(temp_spiking_acc_ls)\n",
    "      val_set_spike_time = np.mean(temp_spike_time_ls)\n",
    "      if val_set_spiking_acc > best_spiking_acc:\n",
    "          # save the model \n",
    "          best_spiking_acc = val_set_spiking_acc\n",
    "          torch.save(model.state_dict(), PATH)\n",
    "      spiking_acc_ls.append(val_set_spiking_acc)\n",
    "      spike_time_ls.append(val_set_spike_time)\n",
    "      \n",
    "\n",
    "time.sleep(0.2)    \n",
    "    \n",
    "    \n",
    "print('\\nFinal loss:', best_spiking_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "pD0QpvrplLxk",
    "outputId": "3748ca5f-2ef5-44a6-d7da-f6c01b0edbd2"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,4))\n",
    "plt.plot(spiking_acc_ls)\n",
    "plt.title('Accuracy plot')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "g5oWzlBqgy6t",
    "outputId": "f33d5ff9-b4c7-4e96-c037-e27083e54b82"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,4))\n",
    "plt.plot(loss_ls)\n",
    "plt.title('Loss plot')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "6pKrRU45g2Qh",
    "outputId": "f042a5b3-f820-450f-c774-2626b9927279"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,4))\n",
    "plt.plot(spike_time_ls)\n",
    "plt.title('Mean Spike time')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Mean spike time')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "id": "LjyK8Mbgg2X9"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "bNVYvBoLs0eI",
    "outputId": "bc85c3c9-2f39-4bfb-8698-77cc0fc21a66"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[[1, 1, 1, 1]],\n",
      "\n",
      "        [[0, 0, 0, 0]],\n",
      "\n",
      "        [[2, 2, 2, 2]]])\n",
      "tensor([[[0.6167, 0.9932, 0.5343, 0.5055]],\n",
      "\n",
      "        [[0.9501, 0.4929, 0.5270, 0.5144]],\n",
      "\n",
      "        [[0.5051, 0.7801, 0.5274, 0.5020]]])\n"
     ]
    }
   ],
   "source": [
    "# unit test for spiking logic\n",
    "\n",
    "############ TEST CASE ############\n",
    "# batch size = 3, seq len = 3, num outputs = 4 \n",
    "out = torch.tensor([[[0.5155, 0.4866, 0.5293, 0.4982],\n",
    "                     [0.6167, 0.9932, 0.5343, 0.5055],\n",
    "                     [0.7138, 0.9959, 0.5277, 0.9173]],\n",
    "\n",
    "                    [[0.9501, 0.4929, 0.5270, 0.5144],\n",
    "                     [0.5108, 0.9840, 0.5448, 0.5071],\n",
    "                     [0.5051, 0.9801, 0.5274, 0.5020]],\n",
    "                    \n",
    "                    [[0.5501, 0.4929, 0.5270, 0.5144],\n",
    "                     [0.5108, 0.7840, 0.5448, 0.5071],\n",
    "                     [0.5051, 0.7801, 0.5274, 0.5020]]])\n",
    "\n",
    "# expected outputs\n",
    "first_spike_idx = torch.tensor([[[1, 1, 1, 1]],\n",
    "\n",
    "                                [[0, 0, 0, 0]],\n",
    "                                \n",
    "                                [[2, 2, 2, 2]]])\n",
    "\n",
    "preds = torch.tensor([[[0.6167, 0.9932, 0.5343, 0.5055]],\n",
    "\n",
    "                      [[0.9501, 0.4929, 0.5270, 0.5144]],\n",
    "                      \n",
    "                      [[0.5051, 0.7801, 0.5274, 0.5020]]])\n",
    "\n",
    "# evaluate outputs\n",
    "seq_len = 3\n",
    "spikes = (out > 0.95) * 1\n",
    "spikes_red = spikes.sum(axis=2)\n",
    "# set seqs with no spikes to spike on last output\n",
    "spikes_red[torch.sum(spikes_red, axis=1) == 0, -1] = 1\n",
    "first_spike_idx_obs = torch.argmax(spikes_red, dim=1, keepdim=True)\n",
    "first_spike_idx_obs = first_spike_idx_obs.repeat(1,4).unsqueeze(1)\n",
    "            \n",
    "preds_obs = torch.gather(out, 1, first_spike_idx_obs)  \n",
    "\n",
    "print(first_spike_idx_obs)\n",
    "print(preds_obs)\n",
    "             "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "456S3jAB1ESr"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "authorship_tag": "ABX9TyM95tGe0n1q0YyaQsBki0vO",
   "include_colab_link": true,
   "name": "rank_coding_mnist.ipynb",
   "provenance": []
  },
  "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}