model_analysis/Oracle_Upper_Bound_Performance.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "f7da718a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format='retina'\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from tqdm.notebook import tqdm"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4896cfe2",
   "metadata": {},
   "source": [
    "## Imagenet V1 MRL Oracle Performance"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8a2b7f58",
   "metadata": {},
   "source": [
    "Just change the dataset and make sure you have the predictions and the ground set ready. This will compute the oracle upper bound"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "cbf61a30",
   "metadata": {},
   "outputs": [],
   "source": [
    "single_head=False\n",
    "multi_head=True\n",
    "fixed_feature=False\n",
    "feat_dim=2048\n",
    "dataset=\"V1\"\n",
    "prediction = torch.load(f\"predictions/predictions_sh={single_head}_mh={multi_head}_ff={fixed_feature}_feat_dim={feat_dim}_dataset={dataset}.pth\")\n",
    "gt = torch.load(f'ground_truth/gt_dataset={dataset}.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "3f5d16bc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Majority Ensemble Performance tensor(0.7677, device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "print(\"Majority Ensemble Performance\", ((torch.mode(prediction, dim=0)[0]).cuda()==gt.cuda()).sum(dim=-1)/len(gt))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "802ea8cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "corr = (prediction.cuda()==gt.cuda())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "c6b692f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "M,N = corr.shape\n",
    "oracle_ = []\n",
    "class_={i:[] for i in range(1000)}\n",
    "tot_corr=0\n",
    "for i in range(N):\n",
    "    smallest_idx = M-1\n",
    "    for j in range(M):\n",
    "        id = M-j-1\n",
    "        if id==M-1:\n",
    "            continue\n",
    "        else:\n",
    "            if corr[id, i]:\n",
    "                smallest_idx=id\n",
    "    if corr[smallest_idx, i]:\n",
    "        tot_corr+=1\n",
    "    oracle_.append(smallest_idx)      \n",
    "    class_[gt[i].item()].append(smallest_idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "418be702",
   "metadata": {},
   "outputs": [],
   "source": [
    "import collections\n",
    "counter=collections.Counter(oracle_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "5eca2235",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "oracle_hist = np.asarray([counter[k] for k in range(9)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "bfef8b63",
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "sns.set_context(\"talk\")\n",
    "sns.set_style(\"whitegrid\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "54ccf343",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[67.462  8.782  2.588  1.358  0.642  0.318  0.204  0.128 18.518]\n",
      "[8, 16, 32, 64, 128, 256, 512, 1024, 2048]\n"
     ]
    }
   ],
   "source": [
    "print(100*oracle_hist/len(gt))\n",
    "print([2**(i+3) for i in range(9)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "e4d58b0c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 263,
       "width": 406
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.bar([i for i in range(9)], 100*oracle_hist/len(gt))\n",
    "plt.xticks([i for i in range(9)], [2**(i+3) for i in range(9)], rotation=45)\n",
    "plt.xlabel(\"Representation size\")\n",
    "plt.ylabel(\"(%) of image\")\n",
    "plt.tight_layout()\n",
    "# plt.savefig(\"oracle_dist.pdf\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "1c8b6cfe",
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_expected_dim(counter, norm):\n",
    "    dim=0\n",
    "    for c in counter.keys():\n",
    "        dim+= (counter[c]/norm)*2**(c+3)\n",
    "    return dim\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "3707a7e9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "391.73904000000005"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "compute_expected_dim(counter, len(gt))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "b61198c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "def calc_acc_from_oracle(oracle_, pred, gt):\n",
    "    corr=0\n",
    "    for i in range(len(gt)):\n",
    "        model_id = oracle_[i]\n",
    "        if gt[i]==pred[model_id][i]:\n",
    "            corr+=1\n",
    "    return corr/len(gt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "1a5f4181",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Oracle Performance on dataset Imagenet V1: 81.542\n"
     ]
    }
   ],
   "source": [
    "acc = calc_acc_from_oracle(oracle_, prediction.cuda(), gt.cuda())\n",
    "print(f\"Oracle Performance on dataset Imagenet {dataset}:\",100*(acc))"
   ]
  }
 ],
 "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}