5_visualisation_poster.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\n10599070\\AppData\\Local\\Temp\\ipykernel_39976\\4104183638.py:39: UserWarning: color is redundantly defined by the 'color' keyword argument and the fmt string \"k--\" (-> color='k'). The keyword argument will take precedence.\n",
      "  plt.plot([-2, 4.25], [4, -1.75], 'k--', label='Decision Boundary', color='black')\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAggAAAGKCAYAAABpbLktAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy89olMNAAAACXBIWXMAAA9hAAAPYQGoP6dpAAB/vklEQVR4nO2ddZgT19eA38i6swIsi7t7aQvUhVIvdfdS968u1N3tV6dutIW6UAVKoRRr0eKwwLLuluT7Y8mylmwyGU3O+zx52k0mMzfJMPedc8691+bxeDwIgiAIgiA0wW50AwRBEARBMB8iCIIgCIIgtEIEQRAEQRCEVoggCIIgCILQChEEQRAEQRBaIYIgCIIgCEIrRBAEQRAEQWiFCIIgCIIgCK1wKn2j2+0mNzeXpKQkbDabmm0SBEEQBEEjPB4PZWVlZGdnY7f7jhMoFoTc3Fy6du2q9O2CIAiCIBjIli1byMnJ8fm6YkFISkoCYOPm70lOTlC6G8EAHn3kTZ584i3unnYpF140xa9BCoJRVLtKjG5CIzUGtaWsVqKzZiYp2piVCmIcKSG9v6y0kv49T2zsx32hWBC8aYXk5ASSkxOV7kbQGbfbzZIlq6iuruHmm57iu+/m8drr99CtW2ejmyYIjVS7ionG+BuPGlcxADHE63bM0iZSkBSr22FDplglmUk1qNMNhWTd21xHjCM15L20Vx4gt44Rht1uZ+asZ3jm2VuIi4vl558WMGLYibw1fRaybpdgBqp3d8pGU6NzO0prbc3kwCwU19oCeuh5PLNhxG9X4yrW/BwVQYhA7HY7l11+KosWf8i4vYdRWlrO+efdwZQTriUvr8Do5gkRjBnkQI8Lb1PMIgZW6YzBvG01ShS0QgQhgunXrwe//vYG991/FVFRTr7/bh4FBebJ+wqRhVnkQC+MFgOzda5qYBZp0Pu31UpqFdcgCOGB0+nk5lsu4IjJE1i5Yj0DB/ZqfK2uro6oqCgDWydECkbLgd5iYAThIgHB0vJz61nj4P2t9apRqHEVq1Kb4EUiCAIAw4f359TTjmj8e968JQzsfwyzZ/9pYKuESCBS5EDvu0oz3EmbESO+Ez1/ezXPZxEEoU3uv+9lNm7M5fBDL+aaqx6isrLK6CYJYUgkyIGenYMIQXDo/X3pdS6odV6LIAht8uFHjzF16skAPPfc+4wZdQp//rnM4FYJ4US4y4FenYFIgTro+T3qcW6oUZcggiC0SWJiPM+9cBtfffMC2dmZrFmziYnjz+HOO56jtrbO6OYJFsdIOdBjlIKeYiCoj16yoJcoKEUEQfDL4YePZ+nyTzn9jMm43W4euP8VPvv0R6ObJVgYo+VAS7S+4FstWpBXbff7sAJ6iYKWKD3vZRSD0C5pacm89faDHHvsQcya9TMnnzLJ6CYJFiXc5UArzCIEanfqwewvK9at6rGDxfsbaDUKQusRD0pGONg8CqfPKy0tJSUlhcLiuTLVcoRSUlLGpVPv5d77rqR3b1m4S/BPuMpBOIqBVe7uwVhx0EoWtBwWGeNIpbS0guz0yZSUlJCcnOxzW+ucBYLpuOWmp/jow+8YNeIkXv7fxzJVs2BKrCgHeqcRrBj692Jk6kKr30jLVFUw/x6sdSYIpuLGm85jv/3HUFFRxWWX3sdRR15Obm6e0c0STIhR0QOt5ECrC7heYmBlIQgEPT+blr+ZdpIQ2Iy54XdmCLrRs2cOP85+hccev4GYmGi++3Yuw4dO4cMPvjG6aYKJCEc5UButxSDchcAfen12rX5DI6fjjqwzRVAdu93ONdeexcJFHzB69CCKiko54/SbefaZd41ummACRA78o5cYCHvQWha0+E2NkgQ5cwRVGDSoN3PmvcWdd11Kly5ZnHb6ZKObJBiMyIF/tBKDSI0UKMFKomDE4l5yBgmqERUVxZ13TWXFqplkZKQB4PF4ePWVGZSVVRjcOiES0EIO1L4wa3GHKVIQGlp+f1aOJsg8CILqJCTEN/7/hx98y9RL7uHhh17j9TfvY+LEUQa2TNALI6IHWsmBmmghBmZhS7kj5H10TXSp0JLQ8H6nag6fVHsOhdJamy4rRJrn7BJUx+1289ijbzJqxEkkJYzDaR9O966HAnDvPS8R5RjB8uVrFe9/+/ZdJMbvxRWX3e9zm+wuWXTvns2GDds46IDzufGGx6murlF8TMH8iBy0Ru2ogRHRgi3lDr8PqxwjULT4jtVOOWiNTJQUxtx5x3M8cP8rdOyYzn77jyYuLpYBA3pyzrnH0r/vUUw6YgIffPhoSMe49uqHefHFj1j2zwz69evR5jalpeXccN1jvP76ZwAMGtSLN996gFGjBoZ0bMF8iBy0Rm0x0AO9O+NQ0CvqoPaETGpFE5REEkpLK+mZdVq7EyWJIIQpVVXVdO54IA6Hg1VrZpGZ2aHxtWuvfphnn32PhYs+YOTI0Drpbdt20rP7JKaceAjvf+BfNr744hemXnwPO3cW4HQ6eeLJG7ns8lNDOr5gLvQWBDPLgdXEwEpS4A+thUFNUTBKEgIVBEkxhCnz5i2lvLySo47er5kcVFZW8dZbXzBkSJ+Q5QCgS5eOHHDgWD7/7Cd27izwu+3RRx/A0uUzmHLiodTX19N/QM+Qjy+YB5GDPaglB1qmEowM32uJ1p9Jzd9ErfNEq3SDCEKY8csvC3Hah3P4oRcD8O47X+G0D8dpH87ttz3LJx//QElJGaecekSb7z/rzFtw2odz/30vt3rtjz+WkpQwjqyM/Vi1akPj86eddgR1dfVMf3Nmu+3LyEjjgw8fZf6C9zj44HGNz//773+43cYuxiIoR+RgD2rKgdqEoxD4Q8vPq5YomFkSRBDCDKfTwdnnHEP37tkAHHnUfpx9zjGcfc4xTDpiPF999RsA+x8wps333z3tMqKinDz15NuUlJQ1Pr927SaOP/ZqAD6f9QwDmtz973/AWAC+/vr3gNpos9kYM2Zw498bNmxl/D5ncdihF7N58/YgPq0gqIOZ5ECLqEEkSYEvtJIFtSRBjXNHbUkQQQgzJkwYxetv3EuHDg15pTfevI/X37iX19+4lwkTRjHn979xOp2MHDmgzff37t2V888/nqKiUp568h0Adu0q5KjJl1NUVMq77z3EvvuOaPaeXr1yyMhIY+GCfxSNUFi+bC1ut5tffl7IiGEnMv3NmbLwk4WwevRAjYuqWhd4NcUg0qIFwaD292KmaIKakiCCEIbU1dWxYsV6unbtRIcOKY3P5+UVsHNnAV27diQuLtbn+2+742Li4mJ55ul32bZtJ8cecxXr1m3huedv5ZhjD2zzPf36d6emppaVK9cH3d5jjj2QRYs/Yu99hlNaWs4F59/JlBOuJS/Pf02DYDwiB+aLGogUBI4WohAqZpIEEYQwZOXKDdTU1DJ8RP9mz+flFQKQlua7ahUgOzuLyy4/lZKSMkaPPIUFfy7nttsv5qKLT/T5Hq+I7NpVpKjNfft259ff3uD+B64iKsrJrJk/M2zIFD7/bLai/QnhRzjLgRoYJQabKxyqP/RGze8unCRBBCEMWbpkNQDDhzcXhJKScgASkxLa3cc1156J3W4nP7+Ic849hmn3XO53++Tkhn2WFJf53c4fDoeDm26+gPkL3mPYsH7k5xcxf/4yxfsTtMWotRbUwAxyoFbUQC8x0LMzN0oc1Pou1fhtzSAJMtVyGLJ0aYMgjGgRQUhJaZivoryddRE8Hg83XP9Y46gCp6P908QrHympSUG3tyXDh/fnjz/f5YXnP2g2T0JNTS0xMdEh71+wHlotwKQUNeQgVLSWAiPu5NujZZu6JWgz54H3uw11ToW8antI8yYU19pCnishlGmZJYIQhngFYViLCEJWVsN8CIWFJX7f/383PsGHH3zLEZMn0rlzJtOnz2Lt2k1+31NUVApAZmaa0mY3IyYmmmuvO7tRCOrr6znwgPO56ooHqKioVOUYgnL0jB6YLbUQznJgZJhfCVq3V42IghkiCUoRQQhDli1dQ3JyIr165TR7PisrnU6dMtiyZSeVlVVtvveZp9/lySfeYuxeQ/jwo0f5v5vOo76+nrvufN7vMVev2khMTDQDB/ZS7XM05ccf/2TBn8t54YUPGTPqFEk9GEgkpxaMlgMt0glWkwJfaPk5rC4JSs97EYQwY8uWHRQUFDN0WF9sttYnxYSJo3C5XCxevKrVa598/D3XX/covXt3ZdYXzxIfH8dFF59Ily5ZfPzR9yxZ0vo9AOvWbaGgoJixew0hNjZG9c8EMGnSeL757iW6dMli7drN7DfhHG6/7Vlqa+s0OZ5gDtSMHoSDHKhFuEiBL7T4fKHKWah1CUZIgghCmOHtxEe0SC94mTx5IgC//vJXs+d/+20R55x9GxkZqXz1zQuN0zPHxsZw080X4PF4uOP259rc56+/LGy2b6049NB9WLJsBmeceSRut5uHHnyVfcadwbJlazQ9rrAHq6YW9Fj5zh+hdAxqRg3CWQp8oYUohIKVJEEEIcxYtrShs2w5xNHLSScfRkpKEh+8/3XjcytWrOOE467G4bDz+axn6NOnW7P3XHjRFLp27cQ3X//OnDl/t9rn++9/Q1SUk3POPVbFT9I2aWnJTH/rAT76+HHS01NZunQ1V1x+v0ysJGhKKBfmUOVADSJRDFqi5ndgdMpBL2Q1xwjkumsf5Zmn3+HPhe8zevSgkPa1detOevUIbDVHtdmxI59rrn6Yu6dd1mzqZ0EbIjV6YGU5iHQp8IdaIyCUjnQIZXRDyKtAVlfIcs9C2+TlFdCvz1Ecdvi+fPTx4yHt69qrH+bFFz9i6fIZ9O/fQ50GhsB99/6PjIw0Lpl6Ups1GIJyrCgIkSoHRorBziplx+4Yp+0yzb5QQxSsJgllpZUMyzm1XUGQeRAikKysdK6/4RzumfYSy5evZejQvor2s337Ll55ZQYXXniCKeRg6dLV3DPtJdxuN7Nm/cwrr95Nly4djW5WWGBFOQgVI+TAKlEDpRKgZJ9ai4P3+wpFFLaUOxRJQijzJKgxR0J7SARBCBvcbjfPPfs+t97yNNXVNaSmJvHsc7dy6mlHSDQhRKwoCEZFD4ySAy3FQAshUIqWwhBqNEHvSIJSQQg0gmCNSglBCAC73c5VV5/BX39/yJgxgykuLuOsM2/htFP/j/x8ZWtECNaUg1AROWiQAu/DTGjZrlC/R6W/o9LzRutJlEQQhLBjwICe/D53OnfdfSlOp5NPPv6eiePPpr6+3uimCToRSvTASnKgxegEM0qBL7SQhVC/03CSBBEEISyJiorijjunMvePtxk0qBfX33AuTqeU3ASLFaMHRs95EAyhyoFamDVaEAxaiIJS9JYErZArphDWjB49iIWLPiQ6Oqrxud9//xu3283++48xsGVCUyIxtWAGObCyEPjC+5nUqFXYXOFQXJegtHBRCVoVLJpLVwRBA2JiohuLFEtKyjj7zFs45KALueH6x6iurjG4debFimsuWCm1oBQ15MDq0YJAUOsz6h1JMFOqQQRBiCjsdjuHHbYvHo+Hp558m73GnMqiRSuMbpZgUfSuO1BLDiIJoyVBCWZJNZijFYKgE0lJCfzvlbv4fNYzdOyYzooV6xm/z1nce89L1NXJwk9eIq32QM8ldY2Sg0iIGvhCjc+u9PvXamnutlD7PBZBECKSo47an6XLZ3DiSYdRX1/PtLtfZOL4cygtLTe6aYJF0LPuQA05EEL/HvSUBDNEEaRIUYhYMjLSeP+DRzjuuIO48ooH6N4jm6SkBKObZTgSPTAXVpSD7VXBdS2d4/QbghxqEaPSwkW9ihbVLFgUQRAiGpvNxqmnHcF++48mOjqqsZixoKCYsrIKevToYnALBTOiV/QgFDnQSwyClYFA96G1NOyscuguCcESylTMamB8DEMQTEB2dhYZGWkAeDweLr/sfkYOP4k33/g8opaStmL0IBT0ih6Ekxxsr3I2e+hxHK0I5XtS8vvoVY+g1nktgiAILaioqCI3N4+ysgouvOAujj/uanbuLDC6WYIP9J4YyQy5YX9oJQdad9aBHFuL4+stCcFi5Plm7jNdEAwgMTGen395nQcfuobo6Ci+/OJXhg+dwqczfjS6aYKKhGP0QAs5MFIM2kKL9uhZp6HnqIZQEUEQhDZwOBzc+H/n8efC9xk+vD/5+UWcfNL1nH3WLRQXlxrdPE2wYnohHKMHZpEDs4lBS8wiCWaNIqghwCIIguCHoUP78sef73LzLRdit9v5/be/jW6SYCGsdLfoxexi0BS126pXJMEq54U1zgJBMJDo6Cjuu/9KjjxqP+rrXaSmNqyf7vF4qK6uIS4u1uAWho4Vp1UOBSV3V5EQPbCKGLRke5VT16GSLdFjVIMRIxokgiAIAbLPPsOZOHFU49+vv/YZI4efxLx5S4xrlAWxanohWPS6S4x0OfCiVjRBr1SDHudHqGkGEQRBUIDL5eKpJ9/mv/82c8B+53Hbrc9QU1NrdLMUEWnRAyWYOXqgBlaXg6YYKQnhhgiCICjA4XDw+9zpnHX20bjdbh5+6DX2GXcGy5atMbppQjvoMXrBStGDcJIDL1b5TMGeJ3oPeRRBEASFpKYm88ab9/HJjCfIyEhj2bI1jBt7Gg8/9Boulz7rwIeK3tGDSEkvBIuS6IHIgbYo+X6NjAL5IhQhFkEQhBA57viDWbp8BscceyB1dfXccftzLF0qkYRwwewTIykl3OUg3D+fHsg3KAgq0LFjOjM+fZK3ps9i69adjBo10OgmtUsk1h6YMb1gRPQgUjrPUEc3KFmvIdgRDXot4qSE8FRjQTAAm83GOecey223X9z43KpVGzjh+GvYunWngS0zB2ZYe0EQrI6eES0RBEHQkMsuvY9ZM39mxLApvPfuV6ZZ+MnK0YNwqz/Qm0iJHniJtM+rJiIIgqAhL7x4O2PGDqa4uIyzz7qVU06+gfz8IqObJQSI1ndrRhUnRhqhSEI4fN9KU2siCIKgIQMG9GTO3Le4e9plOJ1OPp3xI8OHTuGLL34xrE1GRA/MkF4wY/2B3sjdtD6YcTSDEkQQBEFjnE4nt99xCfPmv8OgQb3YubOA44+9mi9m/aJ7W6ycWhCEcMWsYimCIAg6MWrUQBb89QHXXX8Oe40byhGTJxjdJMFihBLujvToQaR/fiWIIAiCjsTGxvDIo9fxy69v4HQ2XLBqamp55OHXqaqq1vTY4RA9kAJFwQjMVoeg10gGEQRBMIDo6KjG/59294vcesvTjB19Kn/99a8mxzNSDsxQfyAIQvCIIAiCwUyYOIpOnTJYtWoD4/c5i2l3v0hdXZ3RzRI0JlwK2YTwRQRBEAxm8uSJLF0+g5NPORyXy8W997zEhH3PZsWKdarsPxxSC0YQrlMsC0KgyL8AQTAB6empvPf+I7z73kOkpSWzaNEKxo4+lRmf/BDSfo2Wg0hKL5i1El0QlCKCIAgm4pRTj2Dp8hkcPmk8TqeDkSGs6WC0HJgJPeZAEIRwQwRBEExGdnYWX371PH8ufJ9evXIan//jj6UBT9UcjnIgIxgEQV9EEATBhNhsNgYM6Nn49+zZfzJx/Nkce8yV7NiR7/e94SgHgiDojwiCIFiAjRu2ER0dxddf/c7woVP45OPv29zOTHIQSfUHghCOiCAIggW44MITWPDX+4wY0Z+CgmJOPeVGzjrzFoqKSoEGMTCTHEQiXRNdQW3fLSG47QXj6Bhnrt8qK9aty3FEEATBIgwZ0pd589/l1tsuwm638/57XzN86Al8/V1oIx2EyKBzXL3RTTCUSP/8ShBBEAQLER0dxT33XsHvc6fTt19XcnN3kb+r2OhmtULSC9pgtjtZQR2CjT7phaxeIQgWo9pVzPAx3Zi78FU++XA2J516SONr5eWVJCbGG9g6QTAfekcPwiV9JBEEQbAILesM4uNjOfu8I7HZGob/5ecXM3LwWdx56/+oqak1qJXhQyh53srSci4cMIlV85eq2KLQkTB78IRD1CY1OrDh0S0RQRAEkxNoAeLnM35le24+Tzz6HvvtfQnLl/6nfePaQIv0gtXmQNi8smGa7Jz+PdvZUtAakSLliCAIggnxSkEwIxMuvORY3v/kPjIyU/n3n/Xst88lPPbwO9TXywVSb7asWk9qVjqJacl+t1MSig71jlY6TGuj1wgGEEEQBNOgRApacvSxE1mw5E2OPm4idXX13H37Kxx2wJX8t3areg31Q7gVJ379zjdcffR1nDDoZI4feBK3n3UXO7bsbLZNcUEJj1z1GMcPOJGbDzyTv779jS0r1zWLHtx88Dl8/vT0Zu/zeDxMO/5yvnnkOV0+SyQSqgwpkbFgpc+sBYoggiAIhtFUCNScwyArK433PrqXl1+/heTkBBb8uYKnHn9ftf1HElvWbeWosydz16u3c+OT11FWXMaTNzzd+Hp1VTU3n3Ir2zbkcuPT13PxnRfyyaOvsXL+kmaC0Kl3N3ZsaC5pS2b/wY71W5hw3qlBt0uiCO0TCZ9Ra2QUgyDohJ4TGdlsNk4/axIT9x/JtDte4f6HL9X8mGaOHqRGexQt2HTJXRcBDXf7bpcbt9vNI1c/3vj6e09/SEVZBU/NepzY+FgAtuys4O27n20mCNm9u7Fi3t+Nf3s8HmY9/y77nzKZwT3S2Fyh9JMpp3NcPdurpAvwhR7RAz1QWqAIIgiCoBlmmNmwa7eOvDr99sa/PR4PF55zP4ccNpZTzziscQSE0Jqqimo+f/FzvvzkV3ZuyaN298iQlA4NdQWuehdfvf0Vp111aqMcAGT16AJATr8mgtCnG7+8/yUejwebzcaS2X+wc8NWrnn5XsXt6xjnYmdVaEtMh6skWCV6EGx6Qc/6AxBBEISQMYMIBMrnn/7Kh+//wIfv/8AXs+bw9PPXk5mZGvJ+zRw9UILb7eayk++kIK+YYy44jm59uxKfGM+bj7xFdWU1AOv+XU95SQX7HLZ3s/c6S/NxRkXRqVfXxuc69+5GbXUNhbl5dMjOaogenHokKZkddP1cbRFukqCGHITD0EY1kBoEQQiAlvUCWtQOtIXb7aG2pp7q6jrq6/fcPbhcbirKa6msrKW6qo6amnrc7vbvLo4+dgJ3TLsAp9PBrM9+Y68R5/LVF3O1/Aim4dZLHmNkxtG88viHrV5bunAV+3SdwgF9T2fD2i0snLOcJX+u5Nn37+KYc49ixPjh9Bnamw2rNtJzYA8ACvMKAUjNSG22r3nf/UGnXjk4o/Z0up17dwNg+/otDdGDjVuZdOFJja8rDU2r1ZFZ5Y7bH53j6g2VAzMWJ4aSXgCJIAhCM/SIBtTU1FNUWEVJcRUdOyeRmhoHQO7WEubP3URVVR1VlXVUV9VRXb3ngnfIpH6MGN0Qvt62pYSP3l3Sat8xMU7i4qPYa59uDBuZDUB5WQ3LluQSFxdFXFwUZ51zLAccNIYrpz7Cin83cMoJt3LWOUfw8BNXkpycEPznsUj04NKbz+CHWXN558XPOfWio0ja/Vk3rcvlmjMbQv1PvXsHPft2ZfmiNTgcdrr06ETh7p/gize/pHBnIT0G9AAgKTUJgK3rttJ/RL+Gfa3ZzNxv/2CvIw9oduz4pARSs9LZsWErcz/7gQNPO4qUjLRm23RLcLG5IviUgRqpBtgjCVaMJqglOHrJgRL0Ti+ACIIQoeiVFiguqmLNql0UFVY2PAqqqKjYM8vh4UcOIHVEgyDU1NSz/r8Cn/vyeDxt/n9TamrqqampbxZtKC6uYt5vG1tte8bp5/Lb77/w/Xe/8Pb0b9iyJY+ZXz+G2+0hKiqwDscqcgDQtWdnjjvjUD558xvefXEmU286ncL8Eq489W7Kist59I1bGLHXQAAGDO2F2+3hsdte4YDJe/PLT8v47cs5APQa2FBb0GdIb1IzUnnmluc498azKSsu463H38FmszFkaI9Wx+/cpxs/vTuLkl2FXPfa/ap+NrUkAayVcrBq5MPMQxubYvP4utK0Q2lpKSkpKRQWzyU5OVHtdgmCamgtAx6Ph4L8SnK3lrBtawmDh3aiW4+Gu8ON6wv55P3W0+0mJESTmhbHmL270rd/JtBwp79hXQGxu+/0vf+NjnFgt9uw222tigo9Hg9ud8OjrtZFVVVD5CE5JY6k5BgACvIrWLRg6+7IRC0lRdWUldU07iOrs5snHnudl169ha7duvL+9L9JSY0lu0sK2TnJZOekkJmVgN3ePCOppxyoMZNica2NvO0FHLvXxTijopgx93luOPdBli9aze2PX86UcyY12/7dl2by6pMf4XF72PeIfRm130jun/oQn674iITd0Ydlfyzn2dteYPum7XTtncNJl57II1c9xn1vT6PjmL2a7e/9+19k9tszmXTRyZx4/fk+26kkigCoJghNMbMoqC0HekYP9ChO9JdeKCutZFjOqZSUlJCc7HsyLxEEIWzQKypQV+cid2sJuVtL2ba1hO25pdQ0SQXstU839juoNwBlpTX8+tN/pHWIp0OHeNI6xJHWIZ6YWGMvvLW1LoqLqigsqKRTdhKJiVE4nU7+XbaDb75YybJly0hPT6dLl4aURlSUg07ZSewzsQfduqfpHjlQSxAAnrr7DaY/9ymp6ckUF5Ry0fWncNktZ/p9b161snKtLeXBd9pKBQG0kQQwlyhoETUwsxyACIIgBIWeIwc8Hg+1tS5iYhoukoUFlbz+0p/NtnFG2emcnUyXnBR69UknOydFt/apicfjYfmy9Rw4fip1dfUcdfSh7DV2X+rrGy4TJ58xgo5dGzrazRtK2bS+lF59U+mck4Ddrs2QSTXXYSiutbFrRyGThp2H2+3mmNMOZtqz1wT0XiWSoEQQwJySAMaJgpaphFAKPc0aPQB1BME8WigIPjBiGGF9vZstm4r4b00+69bk07V7GkceNwiAtA5xdOyUSFp6PNldUuiSk0Jmx9YheCtis9nI6ZrJMcdP5OMPZjNr5nds3ryBR564lsSEVDp0dAMNHdCmdaUs+WsXSxbuIjbOQY/eKfTqm0K3nklERWvXSYWCx+Ph8Tteaxzx4XBo286uiS5FkqC0YBHUrUdoSdOOWmtZ0KO+wOxyoJRQRy94kQiCYDqMmleguqqO9esK+G9NPhvXFVJbu+cfc1JyDBdfsU9ETSz0yUc/cd2VT1JYWEpMTDS3TTuDqVce3dipbtlYxsrlBWxaV0p19Z7vyuG00bV7Eocf04PomNA7KjUjCHfe/Dpvv/A5Ew4Zw6rl6ykuKOGTuS/QvXd2u+/VM80AoUURQNtIgi+USoMRxYZ6ywGYI3oAkmIQLIRZJhp6439/UpBf2fh3QmI0ffpm0LtfBt16pOJ0mvOuWEt2bC/g0ovv54dvFwGw34HDmPHVtGbRErfbQ+6WcjasLWH92hJKS2pJTonm7KmDGoWqYFcVqR1icTiC7+zVEoQ3XpjFPTe/ypBR/Xj58/v57J3vefTWVzjsuIk8/Or/BbQPq0kCGCMKZibUuSP0kgNQv/bAi6QYBNNiBiGoKK9hxT87GTmmS2PHP3hYJ/5dtoM+/TLp0y+DTtlJERUxaEmNq5i0LAfvf3YHb7/+A7ff9BrjJw5plUqx223kdE8ip3sSEw7uQsGuaior6hq/O1e9m0/fW4vDYWPgsHQGDUsnJTVG18/y1WdzuPeW1+jaszNPv3cncfGxTDl7EtOf/ZQfZs7h/KtPpP/QXrq2KRBCSTV40TLlYDXCXQ7URiIIguaYQQigIf+8cX0hy5ZsZ92afNxuD0ceN4iBgzsCDXfCWhXaWQlfIxS2bMqjc5f0RqFau2YrSUnxdOrsf7rg/LwqPv/gP6oq94SRu/VMYvDwDHr2TWk3qhBqBOHPuf9w9nF3kZQUzyc/PExq1y6Nr3342lc8dNNLTDhkDM9+cFdA+9M7igASSQgVNWac1FMOQLvoAUiKQTAYs0gBQG1tPUsW5bJk0TZKS6obn8/uksw+E3vQs3e6ga0zB8EOW6ypqeOQ8dezPbeQx56ZynEnTvC7vcvlZsPaEv5ZUsCWjWWNz8cnONn/0Bz6DEjz+d5QBGHtqs2ceNhN1NXW8+4X9zFybP9mqzrW1dZxzNiL2bEtn9e/fIiRew9ud59KBQGMlwSIPFGIFDkA9QVBUgyCaphJCrxUlNfw5ssLqaqqAyAm1sngoZ0YOqIzmVmRLbahzGVQkF9KVJSTosIyLjjzUb6aOZ9Hnr6EtA5JbW7vcNjpMyCNPgPSKCmuYcXSAlYsK6Cyop4OGbFtvgdCjx70HdCNpZvf9/l6VHQU3yx9I6h9ZsW6Q5IEpaiRboA9HWa4i4Ja61TovYSz1nIQDBJBEELCjFLgdrub5ck/encJpSXVjNu3OwMGZwU8jXC4oNWkRnV19Tz+4Ec88fDHuFxuOnZO49n/XcXBh40K6P0ul4e87RV0ztlz/fj1+y04o+yMGpdFXHyUqiMYmlIc4n6NiiKAepEEL+EmCmquxBiKHJgxteBFUgyCZphRCqBhHYK/F25l2d+5nHnBGBISogGoKK8lLt4ZFvMU+MLINREWLVzDZRc8xX9rtgFw7oWHc/9jFxIbGx3UfkpLannrpX/xeBomnho6MoO+ozoSFx+leptDFQQIL0kA64uC2ks0W0UOQDtBkBSDEBBmlQJomNRo0Z9bWDh/c+Pqh/8s2c648d2BhuGKVsNKiyCNHtuPX/58knvveJv/PfcFa1ZtJTo6+EtLUnIUR07pxYI5O8jbUcniBXks/zuf/iMyGDausypzKpgFpRMoeVEr3dAUK6Ye1JYCCD2lYHY5CAaJIAg+MbMUeNm0oZAfv11DUWEVAB0y4tlnQg/6D8wyxYgEK3X0avD7L8vo1r0j3Xs2jAypqqrBbrcTExN4FMDj8bBpfSnzfttBwc6GeSniEqI48JheZHYOfjnqtlAjggChRREg9EgCaBNN8GJGWdBCCrxEihxIBEFQjBXEwOPx8PXMlaz8dyfQsDri/gf3ZsDgjpqKQaR1+MEy8YBhzf6edtt05vz6Dy++fg1Dhwc2z4DNZqNH7xTSclLYtqGUBb9spba6niQV505IjfaoJgmhEGokAbSJJnhp2RkbJQxaSoEXq8mBHoggCIA1pKApNpuN2DgnNhuMHJPD+P16hrxConT+6lJSXM7MGXPJ21nMIeNv4P9uP5Wrb5gS0IyUpbU2bDbI6ZVC525JFBdUExvX8Pt6PB7Wryyke980nFHG1pWoMaJBLUkAbaMJ0HZHraY06CECLVFjlIIRcqBlasGLpBgiHCuJQe62EmJjo+iQHg9ATXU9xcVVdOzU9tC6thAJ0Jf8XSVcf+WLfPn5HwCM3qsfz796NX375fh9n7/RC5vWFvHLFxtITIlm3EFdyempbOVMNSMIagx7VCPdANpLQrhgpBiAwXJQXUHPrNPaTTGEb1m34JdqV7Fl5KC+3sVP36/lvTf/5odvVuN12phYp085qHEVt/kQ9CUjM4U337+JF1+/luSUBBYtWMOB467l5Re+bFxRsSXtDW10OOzEJ0ZRXlLL7M/W8dvXG6itCf5CreYdmBphYrVW+uuW4NJ97L6VUOv7saocJAfxfhGECMNKYgBQkF/Bu2/8zd8LtwKQnBJLfX3zf1wiAubGZrNx8ukHMOevp9n/oOFUVdXy0D3vkb+rtNW2gcx7kNMrhePOHcSg0VnYbLBhVRFfvL2SvNzyPRu5XMT+vpyEj38l9vfl4LJGh6nmcsAiCs1R8/swSg70RlIMEYCVhMCLx+Nh+ZLt/PT9Wurr3cTFRzHpqAHk9JKyGSvjdrt5/eVvyMhIaTU9s5JJkfJyy/n9642Ul9Zis8HYA3IYveU/OvzfKzhzCxq3q89Op/CRi6g8Zt9W+zBbqsGLWikHL5GaelBbkoyUA7WiB6WllQGlGEQQwhgrigFAXZ2LH75ezYp/GkYodO2RxKFHdSchUf0JcwTj+e7rhbz5+o/c//RlZGSmBv3+2hoX82dvZsOqIk5OLmLA9U+DB5p2+57df+x6++ZWkqD2aAYzS4KXSJAFM4kBmEcOQAQhorGqGEBDuqCuzs3Hb62mML+affbPZtS4rIhedjmcqaurZ9TAS8jdmk96RgoPPHM5hx21d9D78Xg8FG6vYPhBV+LILaCts8VjA1d2Blv/eQUczTvISJQEL+EkC1qlVMJJDiBwQZAahDDCavUF0Lp+ACAqys4Rx/fk+NP6MnrvjiIHYUpprY0qTxSvfHA7/Qd1pyC/hEtOf4Abpj5FaUlFUPuy2Wx0WbcBpw85ALB5wLktn9h5K0JvfDuomWfumuhStTahJd7cvFXrFbRsvxrfvdFyEAoiCGGA1cSgrWLCFcsKWLpoV+PfaR1i6dJNIlPhRmmtrfHhZdCwXsz89QmmXjsFu93OjPd+4oh9rmTer0sD37HHQ9yX8wPa1LGjMNhmK0LtYjQtJcFL087WrMKgRxvVkjIzyEEwoxZaIhVfFscqYuBrZIHH4+HvP/OY90suAFmd4uncRZ3pdM1KKCsUhvKPXW+C+ZwxMVHcNO0cDj5iLDdc8hSbNuzgjKPv4Js/nmHA4B5+3xv72zLS7pxOzN9rAzqWq1OHVs9pNbOi2ktDezstrdMOXtrqgPVKSRghKGpJmBpyaLQcgAiCZbGCGLQ33NDj8TDnp20sWdgQORg1LotO2fE6tEwZWi09bLU2aMmYvQfx1dynefCON6iuqvUrB9HL1pN213TiZi9u9rwH2q5BAOq7ZFC976A292cVSQD9RaEpwXTcTWXCrBEJUDc6Ey5yACIIlsPsYhDoHAQul5vZX29m9b9FAEw4qAsj98rSsGX+CfeO10okJMZx35OX4Woyd0HejkI+ePN7pl43hfhtBaTe9y6JH//a7H21g7tTOWksKU980iAJTa6P3v/9+bgj6emCKJ37VW+nEU6iEAhmlgJQP20TTnIAIgiWwcxiEOzERHW1Lr7+bAObN5Rht8PBk7szYEjrsK/aiARYC8fukQYej4ebr3iOn7//ix//9yXvllSQU7/nwl7fNZOi286g4pT9weGgdmSfVvMg1HbqwBcHHcbOoYPp6ec00HoRJy2iCWB+UTAbZhQDMJccgAiCJTCrHCidsXDVP4Vs3lCGM8rOEcf1oEdvZXPp+0JEILywl1dxVnwMy4DlBaWMAe4Drk5LpPz/TqH0wsnQZDnpymP2pfLIccTOW4FjRyGuTh2o3ncQPfNrGJIcTZTe4YMWaCUJ0LzjE1lojlZFnmaJGoD6NUoiCCbGjGKgxjTGQ0ZmUF/voXOXBDqFWJAoMhDG1NaR9Pp3pD7yIefll3AEcBHwJfB/wMd9uvDY5HF0i2ljAi2Hg+qJQ5s9ld6xeX3LprXFdOmR3GpFSD2WgtYq5dAUkQVtR36YKWoA2hQwy0RJJiQcxcDj8eDxgN0e2oVXhCACcLtJ+OR3Uu97l6iNOxqf9jgdlJ59KC/3yWbaA+9TUV5FfEIsb8y4i732HRzUIf5ZuINFv+fStXcKBxzdq83zUmtJ8KKlJLRFuMuC1sNB1Ry+apQcBDpRkkQQTIbZ5ECthY/+/jOPzRtKOfKEXkTHBH6BEiEIU1yuVikA7HZiZy8m7e7pxCzb0GzzihMmUnTHGdT3zuYkYO+j9uaGS59mZ24BQ4b3DvrwGZ0SsDtsbFlXwoKftzDuoK6GTcilRzShKeEYWdBjjgi157Uwc+TAiwiCSTCTGKi9GuLK5QWN8xysW1PMwKHpPrcVIQh/4mfNa1VE6MpMoT4zhZgVm5ttW3XgcIruPofakX2aPd+1Ryfe/+p+dm4vJD4hFmhYCGruL0uZeNDIdtvQqWsSE4/owa9fbmD10nziE6MYNq5zs230SDU0RW9RgLY7VrNLgx4y0BSzigFoPy+KCIIJMIscaLFM8qb1pfz0TcNFf+ReWW3KgVWlQM/OA4ydclUt4mfNI/Osh/aMO9yNY1cJjl0ljX/XjOhN0d1nU+2ns7fb7XTuktH49/SXvuSem1/lqBMmcs/jl5CW7jt0CtCjXxpVB9ax4OetLJ67nbiEaPoOaX5+6i0JYIwoNMVXB6y3OOgtAi3RYllmK8kBiCAYSjiLAUBRQTXffLYBtxv6D05j/IHZja+ZVQr07gyCIdi2mU4oXC46/N8rrVZabIrHYWfXy9dSOWUi2IPrIKura3E47Hz56e8smPcvDz93JQccNtrvewaOzKKyvI5/Fu7kjx82kZwWQ8cuzWuqjJAEMF4UWmJ0h60XZhcD0G9GVXOceRGIGeSg5XoIauJyufn+i43U1bnp0jWRgyd3o6zO3moefiMorrX5fIQT/j6nEZ81dt4Kv4spAdhcbtydOgQtBwCXXncin85+lN79csjbUch5J07jlqueo7ys0u/7Rk3Ipmf/NDwe2Lm1vM1tjJStrFi3Jp2W0BytvmerygGIIBiC0XKgpRh4mf/7dvJ2VBET62CfST2ocOmf1zRLx2hW9JYHx/aC9jcitMWUho3qy5e/P8kFlx8LwAdvfs/k8VezeOFqn++x2Wzsc2g3Dp3Sh2HjOvnczuiIjLcDE1lQDy2/09Roj+opBb3XYpEUg44YLQagXTqhJZ17pZG8qoTR+3UhISla8+NJx68uvr5PpRc8e14Ryc/NDGjbthZTCobYuBhuf/ACDpm8Fzdc+jTbt+YTHe3/UhcV7SC7u/+aBTAu3dCSph2aWVIQVkFrwdJCJI1apE3mQdAJo+VADzFomTpwudw4HNpcvMxwkRYaaO+CGPvrUjIufALnzoZ1N3wupmQDV3YGW/95BRzqRJzKSitZMPdfDj5ibONzRQWlfgsYy0trmT97M3sdkENyWqzP7cx4DoostI0eUZcEWx1fzvidb2bOY+U/G8jPK8Zut9O1R0fG7z+c0y+YRO++OUHvVws5CHQeBBEEHTBSDvQUA4/HQ1lJLcmpMaofw4wXY8E3qdEecLlIfehDUh75ENvuy4wrJQF7SQXYWiymtPvn3fX2zVQes69m7VqxfAMnHXYTU6+dwqXXnYjT2VpEfpq5ji3rSkjvGM/kU/tjd/g+98x8XkayLOiZhkmN9rB88X/cMPUp1qzcTHS0kyEj+pCdk0lZaQWL/1pNaXEFTqeDZ9+8kUkBnt9aRg1qKqLITp8sEyUZSTiLQVuFhmuXF/Dnz1sYs18XBo4MfWXGSL74Wj3PXL6pkO5THydp7vLG56oOHMGuV68j9o8VredByM6g8OELNZUDgC9n/EZlRTVP3Pcus79ZyOMvX9Pqrm7cQV3ZubWcgp2VLJm/nVHjs33sbU/0xIznalvnULhKg1H/XlKjPSxesIqzj7+L8rIqzrhgElfffBqZHdMat6mrq+fjt3/kobum44wKrMvVUg5iHKnUUBHQthJB0IhwlQNfIxBqquv57PV/qal2MXq/LgwZ0zHofRt9kbXqxdNsMpH08990u/QJovIb5jXw2O1sv+UM8q4+sXF0QqqjvvVMiiqlFfzh8Xj4/MNfuOvGlykrqSAmNpqbpp3DOZccib3JyImNa4r49csN2O02jj13UEBRMaPPXyVY8Zw3+nxvmlLL31XM4XtdQWFBKTdNO4ep107x+b41KzeTlp5EZlaaz21AezkAKC2tCCiCIIKgAUbJgRFi4GXhr1tZsSiP1PRYjj5rYMBrLuh9UbXiBTEUdL2Y1rvo9PC7dHzqk8aUQm3ndDa9fCMV+/hfK0HvEQLbt+Xzf5c9w5yflwCw7/7DeOSFq+nSNbNxmx9mrCV3Uxnd+6VywFG9At63FUWhLYz+t2K0CLSkrXP0ynMf5ctPf+egw8fw2sd3hrR/rQsRvXIAgQuCpBhUxgg5MFIMAMqKa1i1ZBcAo/frEpAc6HERNfoCZwZ8fQdqX3yjcvPpfvGjJM5f0fhc6SGj2fT8tbjS21/Ou+n5oIcsdO6SwVufT+OdV7/hwTveYN6vy/hu1jzO3z08EhrO5dy3V7FpTTF5ueVkZQd2I9S0/VaWBbN10Ebh63xct3YrX302B6fTwbTHLlG8fz1GKDSVg2AQQVCJcIsaBDOZ0d9ztuF2eejcLYkuPXzbqNYXSxGCwGnru1LaIST98BfdL38CZ2EZ0LDqYu5tZ7Pr8uMUTXiklyzYbDbOumgyEw4cwVsvf8W5lx7d+JrH46FDZjx9h6Sz9p8CVv6dF7AgNMXMNQqCbwI57z6c/gMej4eDjxhLTvfgU6pgbjkAEQRVCCc5CHaWw7zccjauaWjHmP27tFoRT8sLowiBurT8PtsVhrp6Ot//Nh2f+7TxqdqcTDa+8n9Ujh2gSpv0kIWefbK565GLGv+uqqzh7OPu5MIrj2PigaNJ7hDLwBGZfvbQPuESVQhngj2/fv9pMQCHTh4X1Pu+/HQOsz6YzbIl66msrGHI0B7cfs9Z7D1+UFD7aY9QxMCLCEKIhFNKQckUyCUF1TgcNnoO6ECHzPjG57W4CJpJCLRYuMZsc937E4aoLXn0uOgREv7aM0NhyaRxbH72alxpSZq0R6/IwhsvzuKv+Sv5a/5Kjj/1QO5+5CIcTvXOPZEF86D0PHK5XKzZvfLokBYrjfojOdrDWy/NpFfvbB5+6hISE2J5763ZHH/EHfww5zGGDOupqD0tUUMOQIoUQyJc5CDUtRHKS2txOGzEJUSpfsHTWwrMvtQtGC8Smd/9wdDrn8JZ3LBugTvKyfa7zmXXJceATd8OTwtRqKmp45mH3uelJz/F7XbTuUsGj7xwFeMPGE5leZ2mM4OKMGiPGudM/q5ixvY+G4C/N7zT7sqhTVMJhQWldGiyvdvtZsLoqxi3z0CefOHykNsWiBxIkaLG6C0HZhQDL4nJ0RTX2qipVWV3mkuBFSTAH/7ar6U82Grr6PfAG3R/dc+UyZXdOrLl1f+jalQ/zY7rDy2iCjExUdx419kcNGks11/8JJs27OCsY+9k/MR9OPqYSZx88YhWqTS1aPkZRBjUQW2RDPT3b6vGoEMLmbDb7Qwc3I1NG3eG1Ca1ogZNEUFQgMgBVJTVUl1ZjyMtQZX2aCUFVpeBYGnr8wYtDS4XaX/+S0xeITVZHSgaN5i4rXkMu+xhUpaubdxs5+Tx/PvoVdSnJEJ1w3NGVr57O1O1OoPR4wby9bxnePCON3nn1a+Z+/sfVFXUsN8RfejcTZs0Skva+iwiDf7RYxRMWockEpPiKC+rYv1/2xjdotMPpvjQ5XKx+K//OPDQkYrbo4UcgAhC0FhdDtSKGixeuIt1S3bSe3gWQyZ2VbQPtaUg0mQgUIKRhqyv5zLgrpeJ3Z7f+FxtWhL2qhqc1Q0hIne0k9V3XsSWc45slVJo+psaJQvFtTbVOon4hFjufWIqhx45jtuv+R8HHHAAq5ft0k0Q2kKkoTlGrLJpt9uZcNBIvp05j4/e+oHR4wYCbYtBfb2LLZvy6Nm7c5v7evXFr9m6ZRcXXHKEorZoJQcgyz0HhZ5yoPaSzKW1NlXkoLjWRkGlh80rGzqQjJzgL5R51XbV5GBLuaPxIQRO0+/N+91lfT2X4Zc8QEwTOQCILiprlIOKHtn8OesJtpx7VLv1Bt7f2YjiUrWXrd7v4JF8NvtxkpOT2fxfMVUVdTz/2Ees/GeDascIBe/Swm09wgWzfb4rbjiJqCgnH739I9Ofm0GCo77Z63V19cycMZcD976WlbsLGlvy14LV3HP7W1x/88kMGtIjqOPHOFI1lQOQIsWA0VsO1EQtMfCyZVUBf/+4kbikaA49awi2ACZGUlMI9GRzhf7isWP1f3zzyPPsWLmGlIw0DjtvCgedeYymx7S5XJx88Dkk7Mhvc6VFAFdcDL8snI4rVfndsxFRBTU7ka/fX82u7RXUO3Zx3x3PEx3t5NrbzuCiq47DocNU0VphdATCqiLz4+e/cuXFz1JdXUvHzmmMHtuPuNgY8vKK+fuvNVSUN+Telq19tdksnQCbN+7k8P3/j30mDOa1d24Mqq4lVDGQIkUVsaocqC0GXjb+0zBrYvfBGe3KgRpioKUUGCEA/qgoKuady24me3B/TnnqXnas+o/3H3yJyuhEhh15aLNtuyWoV5DY8a9/SNyR73cbR1UNySs2ULTvMMXH8Z4PeoqCmrUJ/YdnsGt7Bbb6ZA6eNJbZ3y7k4bumM/ubBTz20jV079V2GNnsWLWDNoKmaYQTTt6PEaP78r/nvuDXn5by8w+Lqa93k9YhkREj+zBh/6Ecd9KEVnJQUlzOqcffS9fuWTz/6tUBy4HWEYOWiCC0Q6TKga87itLCKgp3VGCzQ/eBGT7fH6oYaCEFZpOBtlj0yZdgs3HSw3cQFRdLr71GUbxtB7+9+m4rQWjr8yiVhvhdhQFtF5MX2HbtYUStghq1Cd37prHwl63UVMdw98NXcfgxC7nnplf4a/5KJo+/mlvvO4/Tz5+k2SgHwRj8FR326t2Zh5+8OOB91dbWcfYpD1FVWcPn395HXFz7C4HpLQZeRBD8IHLQmrxNpQBk5iQTmxDV+vUQxEBNKbCCDLTFuj/+ou/4sUTFxTY+N/CQ/fjrky8o2rqdtBz/d6hKpcFeH5hYbEnKYEe5Q9XhlHnVdstIgjPKzr6HdScq2kFWdgInnXkI+0wcyo2XPcP835dz+7UvsujPVTzx8rUqtlowCi2mQr7xqpeY9/u/PPXC5WzauLNxeGNMTBTDRrReFMwoOQARBJ9YUQ60FAMvBbkN8+1ndm2etzJaDKwqBC0p2LyNvhP3bvZcRo+GUSL5m7a0Kwht0fK7aSkMfWb+yL7TnvW7D48NKjpmsnPMEKD5b6aGLOiZeghVErr1SW32d073jrz7xb28+eIXPDLtbSYdu0+ILRSMROv1EX79aRlut5urpjb/N9e1WxZL1rzS+LeRYuBFBKENRA58M2ZSLwq3V5CY2hAWM1IM9JCCnVX6ikd1aRl1sc2LAOOSk3a/Vq7KMbzfW1RVFSc/8Rz9Pv2+8TXvpbHp2eDZ/cf8Wy/B00Yhnvd3tJIoqDkUEhqGvZ1/+bEcNWUiWZ06ND6/8I8V9OmX0+5Me4Jx6LFgUlOaSkBbmEEMvIggtEAvOTBLSiHY6mWHw05mThJ51XbKqoM/nhmlQG8JCISmbaqvafj/klo7O6scdIwLvSPOXLeRk2++j6z1mxqf+/vYSRRPHM5ej73WrGCxomMm82+9hE2HTfC7T6uJQiiSsHNbOZvWFNG5ezJde+1ZzrqpHOzcXsDFp95PVLSTh569goMmjQ25zYI66C0FgWAmMfAigmAAVpUDUB4xCEUM1JQCM8pAU6KTkqgtr2j2XG15Q+QgJqlhOHFbnyFgafB4GDnrOyY//BzRNTUA1MTF8uWtV7Ns8iEA/LbfgXRf/A89ynZRmdmBnWOGtBk58IWa6Qet6xOUSsK2DSWsXLyLmmpXM0FoSmlJBRlZqfy3egsXnHwvp5xzGLc/cD6JSfFtbi9ohxmFwIsZxcCLCEIT9IgemCGloHTM89xvNuCMdZI9tBMxie1X3oJyMVBDCswgA7On3c9/P8xm7IXnMercs5q9tuOff/nq6htwxMRw7IvPkNa9G6ldcyja3HxSleJNDX+ndvM9Y2Ug0hBdWcVRDz7N8K9n72lD3158/OBt5Pfs1vicx+Fg45jhbNz9dzeH8k5ejaiC1tEEJZLQuVsyyxfsZPvmUjweT5ujFvoO6MYXvz3BY/e8w+svzOLD6d8z95clPPbSNYwbP0St5gstMLMMNMXMYuBFZlLcjciBb/Kq7eSWuMlfX8iOf/MCWrFP6eyGmysciuVgZ5Wj2cNotlc56X7mhdidTpZ9+Ak15XtqCEq2bOW7m28HYNLD95HWvaGDzhk3li1/LKB+9909wPpffiUlJ4fkLtlBHb/pd9Fx7XouOfOyZnKwcMqRvPLmM83koC28v0ko0qbGbJdmWu47KzsBh9NGVUU9xQW+c22xcTHc/uAFvPfV/eR0z2LrpjxOm3wb99/2OgrnqBOakBztafUwO3rMgKgWEkFA5MAf3otyVUk1eCA6IYqYBN/L3SqVAqVoKQLbq0L/55GQ3YWuk45k05czmffupww4+3xqiov5/fpbqCktZ8xd9+DpM4LtVQ3bpx9+PK6PP+WHO6Yx9OQTKVizlpUzv2T/W25U1gCPh4kzv+TkZ54jqrYOgKr4eL68/Rr+OfzAoHfn/a2UzrewJcQhklpFE4KNIjicdjpkxbMrt4LigirSMuL8br/3hCF8M+8Z7rv1dT6c/j3VVbUyV0IQWKHjbw+rSEFTIl4QIkEO1Kg1qC5tuKONTYpta3MgeDkIJVKgFmpIQHv0P+Mctv7wLes//YTuk4/mr2l3UJm7jWHX3EDnfSc22zYmNZV9HnqC5c89xTc33kpMWhqDpl5B0v6TGyWic1x9G0dpTWxFBWc88gRjZ//c+Nymfn159Z472ZXTBXbvT0nRYyiioFbawWhJSEqJYVduBeUlga1znpgUz0PPXsERx+7L2H0GNT5fVFBKYnI8UVGReTkOh87fF1aUgqZE5hlpQYyUA4CasgZBiEluXXughxioIQV6yEBbxGZk0OPYE1j30fv8OvV8aktK6HfGOfQ4su31FVL69GXCU8/73F9bn6OlNHRdvYaL7ryHrG25jc/9POU4Zlw+lfro5hGgpt9tsLJgpCgYLQlJKQ3fY1lxTTtbNmf/Q0Y1/r/b7eaysx+msqKaJ16+lt79coLalxkJ5w4/UKwuBl4iWhCsEj3QSw785Xiry7wRhD2CYHYxMEoI2qL3CSex7pMPqS0poethRzDg3AtU3X/jZ/V4mDzzU85/+QWi6hpSCpWJCbx9840sPmC/dvfj/b71FgUzSUKgJO2eC6Syok7xPjasy2XF8vWUFldw5IRruOnuszln6lHY7frVW0iHrg7hIgVNMc8VVGdEDprTXgFYze4UQ8xuQQhGDoIVAzNLQX5Z8L+Hx+Nh7XPPg7uhI6txOxr3k5Gk3sU5obyMKx5/mH3n/Nb43Jr+A5h+7x0UZAc3A6PSqIJSUQglmqB2XUKgUYRufVI55dIUYmKVn6+9++bw3fznuOnyZ/ht9mLuuflVfvj6Tx554WpyumUFvT/p7PXFilIQ60il1hHYdTIil3sOZznQam6DZZ/9S0VBJZmHDCYmI7DlfvUSAzWlQIkAtMfGN58nd9YHpI7am4oNa6kvLWbE028Tl+172KKXQAWi76oV3Hj/3XTcsaPxuZlTTuatCy6hPqr5mhmB1jC0REmtgpKIQii1CWpJgt6rG3o8Ht57/Vvuv+11qiprSEyK486HL+LEMw5us5hRRMA4rCgFXmJ3t720tJwOqePbXe5ZBEEDrCIHwQ4b21QENocNWwDhz2DkQIkYqCEFWshAS3K//JiNrz9DYt+BDL7nGfJ+/JINrz1N+viD6H/9tKD3Z3e5GL5yGT2q8ilKT2fF4KEcNfNTzn71JaLqGzr+sqQknrnhFhbs63/mQzOLQiRKAsDGdbncMPUpFv25in4DuzHrtyeJiWkQPJEC47CyFMAeMfAiguADkYMGgpEDrdIJwYpBKFKghwy0Oua8n1nz+F3Edsxm6IMvEpWShru2hr8vO43aonyGP/YaCT37Bry//ef/ytWvPU3Hgl2Nz9VERRNTt6eKftWgwTx2613s6tgpqLYqkYVgRUHPaIIakhCIICz8dSuVZbWM2DeblA6+R/gEg8vl4pVnPmfiQSPYZ2zD6n6+JmMStMPqUuClpRxA4IJgnplHdEDkoAEt5CCYiXSCnchoe5VTkRzkl9kaH3qTu2Axa5+6D0diCgPveIyolDQA7NExdJlyJng8bHr35YD3t//8X7n/kdvJbCIHQDM5eOe407n47udYGR/8io9KvuNgf0clky0pnVxJjUmVAvl3tW1DKRvXFFMVQqFiSxwOB1OvndIoBwBPPvIJU897kpJidRbsEtrGO4lROMhBrCO1TTkIhogtUjQjZpKDphfm+ooaihdtwB7tpMPefVptG4wYBEOwHZbaIlBSoiykW71tI5ufuRXsdrpd/RC18V2obbKv2L2OImrGuxT/PZ/chUtJ6Des2ftTUpp/DrvLxdWvPY2Hto3eAxQnp/C/My7GvXvNhKbfRTCFkE2/80CjCsEuILW5whFUNEHpKAc9RjjYHQ3fs9utbvi/aTphV14xjz/4EdXVtcz5bTnP/u8qDjxkhKrHi1TCQQRaEqoUNCViBMHs0QOzygGAp85FdW4x9pjmp4sW6QSlkYJQUCoCvojt0oOBz3/t83W7M4p+j38ScHvGrlraLK3QEhuQVlrC8JXLWDxkZKvXQ5WFQEQh2OGReklCqLQ3osFub/huXS51zqG26gwys1L5/Nt7ufzCp1n3Xy4nHnUX519yBHc/cC4JCeqkNSKFcBQCL2qKgZeISDGIHCiXA6ChB2qB2umEYEPcoaQPSko8zR5mJ6O4IKDt0ova307JdxbMbxNM2iHYlIOSdINe6zeoEbvyV4Q4du8B/Pznk1wwdTIAr//vGw7Y6xoWzl+lwpHDk6bpgnBJG7SFGqkEX0SEIGhJ2MsBNC7O5NkdRg3koh5MR6FEDILBakLQkvzU9IC2K0gLbDtQJljB/E7B1iYEihGS4O/fm9vVkMLwphq0JCEhlkeeuoRPvpxG5y7prF+3nRMm30lhQanmxzY7kSIDTdFSDLyEfYpBjzkP9CQYOVCrGNG2O4zqdnkCloNACLTDURolUJPSEu1y2ckp/n+nRf2GsSMtk6yiXW0avRvY2SGL37oMxb37c7esY/BHMJM2aZV2CCbloCTdoFU9grf2IFRBCGYI44GHjGDuome4+bpXGDq8Fx3SfVehhyOR0Pn7Q2spaEpYC0K4pRaMkAMAm2P3vtxucHvA3nY7jBYDNaRASxFQcszkFDtuu4OHTr+KJ56/AzfNw37edz582pW47Xu+/6bfRaCyoKUoWF0SfNUieAeJOxzKoxRK5jdISU3kxdevbbZk9MI/VzPv93+44trjcDiMX+48VCJdBFqipxh4Cet5ELQUBLOnFgIRhICHMJbbsX39Bza3B/fBoyG+dWFUoHUGgaCXGBghA6Ewaelv3PP5s3Qq2lOwuL1DFg+fdiU/jtk/oH0EE1kItKAx0NEOwYx0CFQUgpWEUKIIbQuCh6qKOmJinTicyiRBjQmQqqtr2W/s1axbm8teew/g+deuoVfv4Ie7GoGIgH+0EIOInyjJzNEDS8nB7pSC7adFUFWLZ98hkNZ8qmUj5ECJFFhNCNrC7nax17plZJUWkpfcgVWjRjSLHASKFUTBKpIQKmoIgsfj4f23f+KW61+hvKyK+PgYpj10HuddNMk0EyyJCASHlhEDEYQwih4YLQcA1NZDlKOxYBGsIQZaSYG9oFLxe93p8Sq2ZA/t1TK0hdqiEO6SoLYgqD198pZNeVxx8TPM+XU50FCv8PRLV9IlJ0PV4/hDREA5eqURIloQRA58E4gcqFWIGIgcaCEGakhBKAIQKmoIRLCyEKgoqBlNMFIS1BCETWuL2LCqiJxeKfQZHPgIkqZosb6C2+3mlRe+4p7b36K6upaU1AR+mPMYvftkq3ocEQH10Lu+IFBBCOsiRbMhctCcQORADzEwUgbaoq32BCsNTb+LQGShJMDRD4EWMm6vcrYrCWoXLwZTuKi0YLFpseKu7RVsWltMfGIUKBQELbDb7VxyxdEceOhILrvgKTp0SFKlHkGEQH2MKDwMhrATBLNGD5TUHQSKWpPB+JWDskpsqzdT5XHAkAE+N7OKGJhNCtqjZXuDEQbv96O2KJhREoIh1KGP5SUN62AkpsSo1SRV6dc/h29/eZjysqrGOoTS0koW/LGSQw4f3e77RQi0wexS0BSZKClA1FiIKRgCjR6EPAnSbtqNHHg82LYXEJtfuGdsVwv0lIPSEnfjIxDsBZXNHlZHyWcJ5vsK5DcIZKKlQM4JJct9+0Lp4k7B4P23WVpUA0BiSrTmx1SK0+kgNW1PCvj2G1/jlGPv4eqpz1Ja2va5EykTDemJd1IjK8kBhFkEwayTImmZWgiEkOUA2OlIpKPdjr2+Hmd5BfVJzetO2usI1BSDYFBDBuy7qkLeR3u4M+MUv7fpZwwkshBoREGtaEIgcyYEEkkwU6qhuqqeovyG8yKjozZFp2rjdrtJ65CIzWbjnTd/5NeflvH8q1czfr8hIgQaYDUZaAuJIASAntEDo0Ys+GNnlQPsdmrTUgCIKSxu9nqochDIFMhKowXBYt9V1eqhB2odN5jPHeh3Gmg0oT3aO08CiSQEOi2z1pGEHVvKAEhJjyU+0bwRhKbY7XamPXges76/n27ds9iyOY9jD7+du256h6qqGqObFxZYNVLgi7ARhHCJHgSKXkWJTS/atR3SAIguLGp8Tg05aI9gxSAYjJCBQAmlbcFIUiCiEIjEmU0SAkVJDc/GDQ2CkN0tqZ0tfaPFCIZA2HfiYP5cPJ1zLzgKj8fDc09/zIRxF7Fq5UZD2mN1wk0KmhI2gqAVehYmqll3oLYcANSkNwhCTFEJuN1+L/aB5KfVihoEKwZmFYL2CFUW2kONaIIadQlqSYKWUQSPx4PDYSO7u/XWQYhxpJKUFM9zL93IJzMfomOnDhQWlJKekWp00yxDOEtBU8KiBsGs0QMzE6wcANQnxOOKjsZRW0vZzgpIbTu0qlfUIFgpUBN7gfL9udOV1xs0Hr/J5wm0fsH7ffmrUwikPiGQ2oRA6hJCrUkIhEDrEYKtRRh5UA/2O6SraWYpDIS26gwmTd6HBUveZN3arWRm7nl929Y8uuRk6dc4CxDuMtAWEkHwQ7hGD5TIAQA2GzXpaVQkJmHztH0x1SNqEOgdcej5/Cqfj1BQe7/Bfk49own+CDWSoHaqIVjK3A5dlnlWA39FiOnpKey19+DGvz/56CeGDTiD5576CLfb+tOTK6VplCAS5QBEEEyBJeRgN6t6DmL18LGUpbWeGEYNOfBHsGIQLGpKgFJCbUOgnz2Q7zLQ2gR/qDEM0h9qphoCrUWoKLVWQV+wIxS++XIeNTW13Hzj80w+9Fo2bdyuTcNMhghBaywvCFqlF8w6KZKWtCcH26uczdZiaEoochBo1KA9lOXnjRWC9lDaPrVFwR9qFC/6Qs05EtSguqKO2e/8w68fraSuVt2Jm7RAyfDFV6ffxjMvXE9CQhxzflvCuJHnMf31r1A4K7+pESHwj+UFweqYKXrgj5Z3eo76OpKKC4HQ5cAfgXRgwYqB2aXAF0raHYwo+ENLSdAj1aBWFGHTynw8brDbbURFm0te1MJms3H+Rcfwx6LX2Gf8UMrLq7j8kkc4+fhb2LmjwOjmhYRECYLD0oIQKdEDo1MLLS/gMVWVDF04l14rl1NU7P8uKlQ5aI/A8+7qSIGtqFq1RygE+3nMIAn+sEI9gsftYdO/+QD0GJIJhDapmdaRxlAnP+rVuwvfzn6a+x6aSnR0FN989Qf//rNencbphAhBaITFKAarovaMiUoJRg4AamLjqImJJa6qkuyiHWzJyGnzvVrKQTBioJRQO3El+/ekxQa9H+9nbG90hPc78zfqob2RDu2Ncigp8fgc3RDqyIZQCXVEw85NJVSV1RIV4yC7T5oWTVQNtWZGdDgcXHP9aRxy2F789ONfHHTI2MbX3G43dru57jFFAtTFXL+uxbFi9EBRjtdmI79TFwByCre2uTaDUjkINKXQHkqiBWre4SsllDYE+pm1jib4++21rEfQOoqw8Z+G6EG3gRk4nOa9dGoxbfKQob256tpTGv/etHE7Y0ecy+wfFqp+rGCQCIG2mPcsbwczpheCQa3ogZ51B01ZG98Zl81OUnUFqZUlzV4LRQ78EUg+PVgxMFoI2kNJ+wL5DgL7LvWXBDUmUfKH0smTKktr2Lmp4TzvMTgjpDaEAw/dN53VKzdx7OQbuOaKJ6io0KeeR4RAXywrCGbDqOhBKASbWvCSX2aj3hHF9tSOAHTP39z4mpZy0B5KxMBKKBGFdrcJURL8oZUk+EOtKELLf3u564oByMhJIrFFGsgsqUI9efyZa5h6+QkAvPq/mew75gLmz/tHk2OJEBiHCEITIil6EIocNO4/oxseoGPJLlIqSgyTg0CjBqpGC4oqlT9CJJjPEGg0wf8+fLc52NU11cCIKELvEVmMPqwn/cd0DunYWqPXqozx8bE89tTVfPHdE+R0zWLdf9s47MArufPW/1FTUxvy/kUKzIElBSESplbWOnqgBuVxiWxL64zbZsNZVOZzO6VyEGhKoT1ClgK1O3mV9hesKPh9vZ3vWqkkWDmK0BSbzUZOvw5k5ChfnCkcOfCg0fy5+A3OOGsSbrebJx59j5ee/1TRvkQKzIfNo3D2i9LSUlJSUigsnktycqLa7fKL2eoPgkkvqDXvgdHRAy/RdTU43G52VLddfR+KHPgj0HSCIjFQ4S4/ZNJ8r5fgi0BHQLQ32sHfKAd/6zj4W7/B37oN/kY2+BvV4G+thm4J7Y9WCGREQ0JtFdGxznbnPEhVuDKjmis66hU98MUXM3/npec/Zcash4iNjQnoPSIDxlBaWk6H1PGUlJSQnOx7wTHz36bqhF7pBavi626vNirGlHIQdNRAxRSAKihoS6CfOZSUgxaRBC1QI4rg8Xj4+4cNzH7nH/K3+o6QQWTWIbTk6GMn8uV3TzTKQX19Pdde+ST/rd3aaluJFFgDywmC2dILRhQnGhE98EXTC39aTSldKvPafY8echAwZpKCtlAgLmaVBF8oTTVoPQ1z4cYiCrdXUF/rIiE1sDviSKfp6pbPPPkRr7z0OfuOuYCXX/wMj8cjYmAxLCcIVsbMdxnBphaguRxkVhdxyI6/GFOwiihXnaKOIlQ5CCpqYHYxaIsg2hzId6GVJPjCbFEEf6LtdrnZtLDhzrf3yE7EJba9tLngm5NOOZj9DxxFZWU11131FMcfeTNbt+40ullCEIggmIxQixPVnhQp0IltdsWkUhKVQLS7nqE7V7c5eRIoHzYXiBwEhEpiUFGdp/gRMkGKgj+0WI8iHKIIWxZto6a0hqg4J31GdtTkGGpidP1BW3Tt1pEffniNp56+idjYGH78YT7Dh07hvXfDc+GncEQEAX2KE9VC6UQv/gg1tQCAzcZf6QNwY6N3XR596lrfKShNLagqBwpRs5NXbV86SILVowhKKMktJXfZDgB67ts9bBdl0ppYRyp2u50rrjydv/7+kLF7DaGkpIyzz7qVu+583ujmCQFgKUEwW/1BMOiRXjAqeuClMCaFxbHdAdinci1Jrj2di6FyoCBqoPpdv1bHCvCz6S0JekYR/BFsmqGuup7/fmlYkCirfybpPTsEfCwzpxD1pK06gwEDevL7nOlMu+dykpISOPW0I4xpnBAUlhKEcCfU4kQlqBI92E1piZulMd3Z4UghGhcHVK7A5mlnUSY95CAI9BACTdpgoCT4wpckqB1FUDPNYHfYSM1JITYllh57dwWsMSeJWfBXgOh0Ornt9otZt+EbBg3q3fj855/NprS0XIfWCcES8Wd+JAxv1CN64MVjs/FLwkBqcJDlKqNXXZ6iULTecmAGMWhJ0G1SQRKUEMp0zC0JZTEnNXBEOei9X0+GHjsQR5SkFoIh0NEJHTqkNP7//PnLOPmkGxg5/CR+/nmBRi0TlBLxgqAULSZHCgWtV7Lz4i964KXCHsuc+AHMievHuijfBV5K7kpBXTkwoxi0JKg2higJZo4iaJVmAKirrmtWOOeMVj6To96YoUAxlKGL3bt3ZtOmXA49+CKuu/ZRqqqstUZKOCOCYBKskl4IlI3RmayOycZe2HanojS10K4cBFFvYHYxaImakuAPJSMbjI4ihJJm2FTo4d+vVrPyuzXUVtYp3k+kEooc7L33MP5e8jEXXTQFgGeefoexo09l4UJtFn4SgsMygmDlAkUjUTO9EEj0oC2iqWe8bQPR+J4210tIw+4MihqUV25v96EWakmC0lSD0VEEtfG43OT/vpqqoioqC6t8Ds8V2kaNSY+SkhJ48X93MuvL5+jcOZNVqzYwYd+zefih10JvoBASlhEEwTd6pReCxXtXub99Hf3suzjEvgYHbm1TC+0Qqhgo7fzVlAY9JEFJqkHNKIIv1Ix6edweCuf/R+2uMhxRDgYe3o/oBJkQKVDUnhFx8uSJLF0+g1NOnYTL5SItzfcaAYI+RLQg6FGgaNahT1qmF1qy0N2VGo+DjrZy9revw4avSZRCTC20QyhyoHYkINR9Gh1J0Bq10wwtJdrj8VC8eCNVWwrBbqP/oX1I8LMQlYxkaI5W0yV36JDCu+89zPc/vsxFF5/Y+PymTbm4XO0vriWoi5z1ClB7giQj6g98oVZ6oendZDHxzHb3pd5jo7utiL2Sc8GHJChCQznQQgzUOobWNRRq1iKYLc1QtjKXirUNk3l12Ls3Kdnq3K2a9YZATfRYS+Ggg8Y1rutQWlrOgfufz8EHXciGDa0XfhK0QwQhjNF6MZtg2Ekyv7l74/HAgPhChibsavZ6SNGDdlDSkeohBmocM6DPpkEUQWmayAy4qusoW5kLQMrI7sR3y9BVwoXgWLx4FYWFJcz5/W9GDj+J1179VKZq1gkRBIujpP5A6/SCr7vITXRgQVlnAEYl7aRvXGHoB1N5wSUjxKCtNgSDGpLgCy3WagiVUM9fR2wUmQcNInloV5L6d1apVcag9xBHI1Zi3H//Mfy95GMmTBxFeXkll1w8jWOPuZLt23e1/2YhJEQQhEbUHr3QEvuuKlZVZrC0PItKl5PttYkNzyuNHqicWjBaDJqiiST4Qc0oglppBrUnTaqvqGn8/+i0BJIHd/G5bVlxGR8+9xGXT7qKKYNP4ejexzNp2HlcdtKdzJ29SNV2Ce3Tq1cOs396lYcfuY7o6Ci+/up3hg+dwscffWd008IaEQRBd5aUd2RWfl/KXU0rxtUPGVptnoOWqC4sYRRF8EWbaTWPB9uKjez8Zim1Be1P6Zu/PZ/LJ13F6w9Np7qqhpETRzDu0L3I6dGJ5X+tprzUYsuEq4QR0YOmOBwOrr/hHBYu+oCRIwdQWFjCRx99J+kGDbHEdGFWnQNBjYKlcM2N1nj2nHo5SRX0SS1hztZO1Hv2OKueFfZmih40pbxyO4nxgYXBK6rzSIjN0rhFFsPtwbbsP2xb8vAANQVlRKcn+n3L24+/y86teVzxwGUcffaRjc9nxbpxu924XMEvRCWox+DBfZj7xzs8/th0LrzohMZiRpfLhcMRntdLo4jYCIJZlnjWaviUUQWKvsLLvsLRTpuLfbJ30jW5kkN6bCPaEeBQpnbuhq2aWmgLVdvn53sz65BHxdS7sP21slEO3MP7kNSvfdnatGYzACPGD2/1mt1uJyrKEvdVqmJ09KAl0dFR3HLrhWRm7llt8+yzbuXyS++jvDwyIzxaELGCEA4YOUGSkuV828KdX8vPmztTU28nM76GST23khCl33S3ZpcDL4G2U++0ipGTJvmlpg7b/H+x7SzCY7fjGTsAuvleF6QpYw4cDcDlk67ijnPu5rFrn+Dtx9/VsrVCiCxevJIPP/iW//3vY0aPPJm5cxcb3aSwQAQhwgi2AlyPcer5VXF8tzGHijonKTF1HNlrC9mJFZofVwgcs9Uh+D2Pq2qw/bYEW1EZnigHnr0HQ6f0gPd92lWncOx5R1NbXcuC2Qv54ePZbF0fuePvzRY9aIuRIwfy/Y8v061bZ9at28IB+53HzTc9SU1NrdFNszQiCAJg/DK7JTXRfLs+h/zKGGKcbg7uvp1h3Upos3hRpfSCVaIH4YRakSe/xEZDcgKehFg844dBeuCTIOVty+OaY67n379WcN/b05jx74d8t/Urbnn+Jg0brAwzrOJoJg46aByLl37MOeceg8fj4bFH32TvvU5nyZJVRjfNsoggCKahst7JdxtzWF3YsF68w+4Bwn9mukAxa5pBDUId6mirr8dWv3sxMJsNz8h+eCYOhyTf0ye3xQOXPcyOzTt5+IMHGHPAaBJT/Bc0hjtWiB40JSUliddev5dPP3uKrKwOLF++lpOmXEd9ffsLxQmtEUEQTIXbY2PB9kx++jedJZtSGp+322Qok2qEWaFiVHEJGfMXkbx63Z4no50QZDHhrtx8Vi5aRc+BPSJeDKzOMcceyNLlMzhhyiG88NIdOJ2RV1iqBiIIginJLYrD42m4e7TbPBw2LI8xvYp2RxUEAXB7SFy/ifRFS3FW1xBdXIKtTnmBq2d3OmvN0rWsX7Gh1esrl/5HVYX1BCpSyczswEcfP86hh+7T+Nw7b3/B88+9j9stQ1UDQbRKMD2dU6vJSKojI6mOLh2q+WtpHNvyZFneSCahtJiMDauJKm8oZq3qlEXJgD54nE5A2ap/WdmZDNt7KMvmL+fySVcxaOxAMjqlU15SQe7G7RTsyGfupo9U/BTmxmrphfbIzc3jyisepKysgpkzf+a11++ha9dORjfL1EgEQTA924rimP1PBhU1DpJiXRw4rpz9xpQRHyfLv0Yajvo6uq1dSf/lfxNVXoE7yknx4P4UDxmwWw5C4+437uSUy08ip08Oa5f+x5yv57Fh5UY6devI1XedKxPxWJhOnTK4/4GriIuL5afZfzJi2Im88/YXMhOjHySCIFiC7cWxfLGoI0O7lTIwu5xunevIzixh2Zo4Vq6PbUxHCOGNBxvJxQUAVGZ3orRPTzzRUartPyEpnvNvOZfzbzm31WtZsRKWtjJ2u53LLj+VQw7dm3PPuZ0Ffy7n3HNu5/PPf+LFl+5oNumS0IBEEATLUO+2s3hjKl/9lszOAidOJ3TpWIfcAIQ3MZUVeH9kt9PJpj4DWT10NCWD+qkqB8Iewi290JR+/Xrw2+9vcu99VxIV5eTzz35i5PCTKC1tf52OSEMEQbAcJWVOfpiXxLzFCSxYHo93KGRMtJukBEk7hIInLdboJjQSU1fNiLxVDFr8Jx127Wh8viwtnYrkFD/vDH9kDoTQcDqd3HLrhfzx57sMGdKHM848iuRkGbnSEkkxCBbFxvqtMc2eGdK3iv49a/hvk52/V7opCbMbgkAXbWp3waa04OYG0IOUlD0popjaanru2kSXwlwcnoawfkJZKYVZez5/5zhzjmtPjZZwlpUYMWIAfy58v9lza9ZsZMuWnRx88DiDWmUeRBCEMMFDXIwHuw369bDTp7uNdZs9/L3STXGZ0W0TvCSn+A5aNojBRnIKc7HvTimUJaewvWsvylPT9GqibiSbVCbCOb3QFjExe0ZE1dfXN9YnXHnl6dz/4FXEx8cZ2DpjkRSDAEBGUtsXq6Z3dnriN9Td5h2wjTl/J/L1b8ls2RGF3Wajb3c7Jx/u4KC97KQmtX5HoHfkArjTg7tIutODj1IM3rqSbgXbsHs8FCaksrDXSNYOGaWaHHST9JPQDnV19YwcMQCAZ599jzGjTmHBguUGt8o4RBAsjJILnllDs2pRWOLk14VJfPJDPRu2ubHtFoWBvax9qptVZtyZCu+uPB7SakqJq98z8dCGrB67xWAUf/UeTVFiB7DJ6BRBP+LiYnn+xdv58usXyM7OZM2aTUwcfw533vEctbX6rTJrFqx91QwBpUU+Zg0L6o2vULGvO0dfHUmwd6aBUl2dxffz3HzyfT3rtrhZunrPELXsTBvD+9mIjTFvx9uUYNoYSv2BHgWKTnc9vcq2csj2hRyy4y/6lO1ZJbEoMW23GCiPGHSUuTEEFZg0aTxLls3g1NOOwOVy8cD9rzB+nzPJzbXeOiehYAlBCOecWHtjq7smKrvghcOFMvg0Q2sKSuDH+W4qm8yQO6yfjb2HOzjzKAeH7G2nf4/sEFuqHarKgVF4PGTWlzKhchVHb53D6MI1pNWV48LeWGtgNO39OzPTHAhajmAI52ttsHTokMI77z7E+x88QocOKdjtdjIzw68Wxh9SpKghqdEeimslRGoUCbFZba5suGGbh9gYDx3TbfTuaqN3V5g4Oodlq0tYuaGSsgrry1WbKBy9EFKUx+NhcvkSOrtKGp8qjYpnXWIXNiV0os4R5bPOxVddjCDoyUknH86EiaOorKwmKqph3o3a2jpyc/Po0aOLwa3TFktEEITwRnEHpLDDW73Rw+c/ufjk+3r+/c9NTZ2HlEQbE0enMnmieWZT0zN6oCS90FbaKJkqBiUUYNs9PBGbjUJHIvXY2ZjQiZ86juK7zuP4L7krdQ5lkxyZtY4mmCGOkqq0Fp07Z9K7d9fGv++Z9hLDh07hlZc/CeupmiWCEIF0jqtne1Xrnz4jyUN+Weu7uZQUGyUlrf8RJKfYKS1pHXp1p8djL2i9pLA7Mw77rqqg2upJiw1pCWJfUQRoSD/MWexm/jLolWOjf087m7fv6bTiYuycfmRHNmyt4r8tVWzdUYNbp2uB6rURGs19YMNDJuV0sxXT1VZEqq0aqiHfmcQOZyoAS2K783dsD2LTYvzvTENkBEP7SHohMFwuF/PnL6WioopLp97LzJk/88qrd9O5c6bRTVMdEQSL0y3BxeYK6y8g406Pw14QnDwADR1fUWsZaYo/SQCod8GaTR7WbGroRLydc/fOJaQmORk5MImRA5OornWzYWsVW3bUsC2vhsIS9e9klYiBltEDX9GdpEwYYltPV1sxcbY934PLY2N7VCpu9ohmtb1hnLnWJZDhUHcjmB+Hw8H3P7zM00+9w+23Pcu338xh+NApPPf8rZx8yiSjm6cqkmKwAOFUqKhkWFy74e8A7o6VdKLb8lL4do6L5WvLqaxyERttZ2CvBA7btwPnHdeZvt32fJaYKBtRTuX1JonxnbWTgxCiB3bcZEZVMDh+F52j90xN6cRNP3s+cbZ6ajwO1rk78Iu7N++mjue7xOHkOZtPhexr1IsZ6g+kQFEIFrvdzrXXnc2Cv95n1KiBFBaWcPppN3H6af9HQUGx0c1TDYkgKCA52kOpisWHWbFu8qrN4Wpapxn8oTiKECDtRRJaUu+CTds9bNoeyx9LPCQl7KRnThzZmdF0zogmd1dN47YjBiSy74gUdhXVkZtXQ15hHUWldRSW1lNV7buDCSWVoIYcNJUvGx7S42pIiq4lLbaWzORa0qOqcNgafve1lWlsr22Yr76QeJa7O7HNk8JOknDvvtdw27S9pJi1/kCITAYP7sPcP97m/vte4cEHXuWrL3/jvvuvIj091eimqUJEC0KMI5UaV7GmxzDrSAZfdQhmpd1ahABSDRC8JDQeHyitSGfpali6GiqqtjVbRTItJQq73UbH9Gg6pkc3e291jZvps3ZQXtlwp9q1U2ccdigphzqF/Z3StILD7iYp1kVSXB3JcfVURdWyrjgZALvNw6SeW1vNTVTlcrKrLp6ddQnAnijQX55uzbbzNQeGv+mV1cJftEyP+gOrFyhK/YFyoqKiuHvaZRx51H5s2phLr145ja/V1dU1jnywIpbpIWIdqVRr3JlbFX91CB3jXOysUqdGQY9iRX9RBKMlodk+4prf+c/5GxavrKdjuo2O6TbSkiElyUZSPEQ5bdjIInF3/zlyoJ2+3Ro6zXqXh5paqK6FmhqorvXw059uXLu/zo7pEB9rw+MBt6dh1eOYqFSgFputYR1Lm63hEeX0EBPtpqrGzsZtMY3Rg0nDdxIX7SbG6cbpaP777azYIwguj538qhhcbjvFtjh21cWTV5dAuSsK74qZimdObAMrpBcEIVDGjh3C2LFDGv/++ecFXHrJvbz6+jQmTBhlYMuUYxlBiHS6JrrYUq5PMaKvNIOaKBnRAOaShJZUVMH6rR7Wb93TwTnskBDfEIHwUl8PVTUe4mJsOB02nHGQsLvfdbtplAOAYf3s9MppeQfuf/WpvAInGyv3TOgSH+0iPmbPTmvrbZTWRlNaG01BVfORBd9u6Kpo2Kke0QMj0gtSfyAo5Z5pL/Hff5s5cP/zue76c5h2z2XExho3kkcJIgg6EEiawYg6BDXTDKaqRQhCEgDVRaEpLjeUtlh2+rdFbn5bBFFOiImG2GiIiW6Y+jmqxc9RUg7bd3lwOqKw2TyN0QJvesOzO6rgwUZ9PdTU2imub17U+fvqdNxuGzV1dmrq7dQmxwHBC6BZowdWSi+YEUkvaMPnM5/m+msf5c03Z/L4Y2/y7Te/8+Zb9zNy5ECjmxYwNo/CWR5KS0tJSUmhsHguycmJarerTbRKMSitQwimUDGQOoT2BKG9CIK/4Y6+0gz+BMFXFKGtNAPQpiAAfgXBXxTBnyQENDdCAJLQFC1FQSlB1xoEUZTYFv6iB77X0/B9TLVGL/iLHigVhEDSC4FEEPSqP9AqgiCCoC1fzPqFSy6eRl5eIU6nkzvuvISbbj4fp9O4+/PS0nI6pI6npKSE5ORkn9uZo3ReCAgt8qVqhm2DXcAJ/N+V+uusApr5L8jhfQmxWaZZzyDotqTFGyIH/ghWDsyGpBcENTj6mANYunwGx59wMPX19dx15/N88/Uco5sVEJZKMYR7oaKZhjtC8EMe/aEk1dAeAc2y6O00g4gmNO2Y9YwqKJaTAERIKzlQs/ZAz+iBWkh6QQiEzMwOfPTx47z37lf88vNCjjp6f6ObFBDm6Y0sSDDhQj0uJP4uiP4upEqiCL7uApV0DEqjCBDEGgIKJwvy3slrEVloum9F+w8gagD6y4E/zDByAYwZvWDG4Y2CfthsNs448yheeW0att1jiQsKijnj9JvYvHm7wa1rGxEErBW+0/vCpuTCrWeqARo6wIBTDiHMKtiyQw+mY1f6Pp8E+DlCkQP/71Ov7sAfRkUPJL0g6MGNNzzOhx98y4hhJ/LW9FmmW/hJBMFkhHphMkMUwR9aSQIEGU1QcfEiX+KgSfQhiLaHKgdq1h34w6rLOkt6QQiVW269kL33GU5paTnnn3cHU064lry8AqOb1YjlBMFsJ7URaQYrRxHAJJIAqouCpgQpBlrKgZqpBX9oFT0wYu0FSS8IbdG3b3d+/e0N7n/gKqKinMya+TPDh07h889mG900wIKCoBVmCuOFSxRBK0lQLeXgxdv5mk0WFLQrkM+tlRwoSS1YNXqgF2a6Lgna4HA4uOnmC5i/4D2GDu3Lrl1FnDjlOl59ZYbRTRNB0JtwjCLoLQkN7w0smhCUKIDxsqDw+IF+Vr3lwB/+zimjogeBIukFQW2GD+/P/AXv8X83nU/37tmceNKhRjdJBEENtAgfmjGKoFQSlKKGJECQaYemNO2stRKGEI8RjBgYIQdWmfPAi6QXBCOJiYnmgQevZtk/M0hNbZjAyOPx8MrLn1BZqd1Kt76wpCBoZb9WC+eFcjekhST4QmkUAQKTBM2iCS1p2Zmr8VBIMJ8nEDHQWw4ketA+VrseCeqRkLDn39ybb3zOpVPvZcyoU/jzz2W6tsOSgmBGtChW1DKKoAWhpBpCSTc07CPwaIIqsmAQwYpBqCMVrCIHahAp0QNJL1iLrt0606VLFmvWbGLi+HO44/bnqK2t0+XYIggWp727IjOlGtrLU7cnCWpFE7xYRRaCbWeg34OV5KA9wi16IAheDjlkb5Ysm8HpZ0zG7Xbz4AOvsO/eZ/DPP2s1P7ZlBcHqaQa9ogjtYSVJgMCjCcFOAGQ2WVDankDFQCs50AqtUwtmix7IwkxCU9LSknnr7Qf58KPHSE9PZcmS1ew15jTeeP0zTY9rWUEwI0aFE0OJIrSHWSVBK1GA5p2zHsLQ8nhKjqlm1CAUObBiaiFQJHogGM2UEw9l6fIZHHX0/tTV1dO3X3dNj2ep5Z7bwspLQENgy0BD6EtBg7LloMH/ktDge1lo8L00NPheHtpLIIs7+VsuuvX+tKkCDmTpaS1EIxj5CUyo/IuZGeVAz+iB1Zd1BokghAsej4e//vqXsWOHND7377//MWhQ78Z1Hvwhyz0bhFmjCO0RygRKoUQSQilehMCjCQ37i1McVfBHWxEAraIQwX6GwKMtyuUgJcVm2boDLeTArIgchA82m62ZHKxevZG99zqDo468nNxc9VagFUHwgRVrEUJNNRghCRB6ygGCE4WGfWojC1qgpK3BiEGocuCPUOTALKmFYDFr9EAIX5YuWYXb7ea7b+cyfOgUPvzgG1X2a3lBMKMVB3uB0LNgUat6BNBeErQQhYZ9x5lOGJS2J7iISvtiYFY5kOhBcJjxOimox8mnTGLhog8YPXoQRUWlnHH6zZx26o0UFBSHtF/LC4KWWNHmtUw1gPaSoLYoKFmRsKUwaC0NahwvWDHQst4AIlcOJHogGMWgQb2ZM+8t7rhzKg6Hg48/+p7hQ6fw9de/K95nWAiCGe3YyCiClqkGCF0SQo0mQOCiAMqiCq2P17oTV+uhuE1BSlCgYmBlORBaY8bro6ANUVFR3HX3pcz9420GDOjJjh35LFzwj+L9WX4UgxetRjOA8hENENyohkBHNED7oxqg/ZEN/kY1gP+RDRDa6AbwP8LBS3sjHbwEMuKh2fZBjH4wE8oiIoGJVKhiAMbLgUQPWiOCEJlUVVXzwvMfctXVpxMVFQVATU0tMTHRMorBiuid6zQykgDqRRMguIgCNL/7DjW6oDVK2xnod6JW1MDogkSryoGWiBxELnFxsVx/wzmNclBbW8eEfc/ixhsep7q6JqB9hE0EAcIjigDqzY0Aoc+PANpHEkDdaAIEH1Fo9X4DIwyhCEswkhSofIWSUgB15ECtqZTNKAgSPRD04LNPf+SkE68HoH//HqxevbHdCIIIQoCYURDAOpIAxogChC4LrfYXojxoEbEwQgxA5ABCjx7ItMqCXnzxxS9Mvfgedu4sAIgsQQCRBF/oIQmgXzQBghcFUF8WjCQYKQB9xQAiQw5AogeCtcjPL+Lii6Yxa+bPIghqEooggLGpBjCPJIC6ogCRIQvBCoGXYBZXspIcgPoLMUlqQYgESkrKSE+b0K4gtH8ltxixjlTNJCHGkRqyJARDarQnIEnIinUHJAldE13tSkK3BJdfSfBe2P2JQue4+nYlISPJ064keDurQETB2wkGIwptdbhmkgalQuDFjGIAIgeCYDSBrNcAYSgIYF5JSI72BB1FCBQ9JQEaLvLtSQL4jyZ4OxwtRAGURRV8dcpaikOoItCUYJdiDkQMQOSgJWatOwCJHgjqEZaCYGaClYRAowhgPkkA9aIJEJwogLKogi/U7MTVJlgpAP3FANSbBMloOQgVmTFRsAphKwhmjSJAZEoCqBNNAOWi4EUNYTASJUIAgUsBBCYGoG/UAEKfSrwlehclao1EDwQ1CVtB0Bqz1iOAMZIA7RcwBhpNgOBEAYIraLSaMCgVAi9GiQEYIwdqLGrmCzOnFgRBbcJaELSMIoSKknoEIyUB2h/hoFY0AYITBQg+qtAUXx2wEeIQqgx4CUYKIHAxAPWiBmCsHIRT3QFI9EBQn7AWBK3RO9UAxkkCqB9NAO1EAZTJQlMC7awDEQm1On5/BCsFoL4YgLpyEExKwcx1B1ojciBoQdgLgtZRBJGEtgkkmgCBpR2geUemRBYgdGHwhR6dvy+USAGYXwzAXHJg9uiBIGhB2AsCmDvVAOaSBGh/QqVgUg6gXjTBS7BRBS96CYOWKBUCCE4KIHAxAJEDf0hqQbAqESEIWqN3waIXLSQBjI0mQPCiAMHLArTd2ZpJGkKRgaaYQQzA+GJEkQNBCI6IEYRwTDWAOSQB1IsmQPCiAKHLghd/nbLa8qCWALRFsFIA1hEDsIYcCILViRhBAJEE0EYSILhoAgQnCmCMLLREyw49VJQIgRejxQDCUw4keiBYnYgSBD2wiiRAYIs8BVqXAIFHEyA4UQBlUQVo3XGqKQxGEooQQHBS4EXkIHBEDoRwIOIEwewFi6CPJIDx0QRQLgoQvCxA2x2r2aUhVBnwokQKwBxiACIHgqA3EScIYP5UA5hXEkD9aAIELwoQuix48dcB6ykPaolAU/SQAghODEDbqAGEtxxI9EDQi4gUBD0IV0mA4KMJoK0ogHqy0BItOm2tUSoFELwYgPWjBiByIAhtEbGCoEeqwWqSAIHVJUBw0QRQLgoQmiyAusJgRkIRAi9aiwGEtxzogciBoDfhfeVsB5GE1iiJJoB2ogChyQK03YFaWRrUEAJQJgVgPjEA4+VA6g6EcMS6V0mViARJADSNJkBwaQdQJgoQuix4sYI0qCUCTdFLCkDZ0swiB20j0QPBCMx1RQxj1JIEIGyiCaBcFEA9WfASSIespkRoIQBtoVQKIDzFAEQOBCEQRBDQb+ijWlMy651ygOCjCaBMFMAcsuALvTr1UAhFCLwoEQMQOVAbkQPBSEQQdhNJkgDBpRxAP1GA0KIK0LqD1FIYzIAaQgD6SgEoEwMQORAEvRBBaEKkSAIoiyZA8GkHCF0UQLksQNsdqFWlQS0Z8KJUCiAyxQCkIFGIHEQQWhBpkgD6RBNAuSiAerLgxV9HawZ5UFsEmhKKFIByMQCRg0CR6IFgBkQQ2sCKkgDKihchtGgCKBcFCF0WQB1haEqwnXN7QqFlZx8ooUoBWE8MQORAEEJBBMEHVpMEMCaaAMpFAUKLKnhRO7oQLGYQgJaoIQRejBADMI8c6JlSEDkQzIQIgh8iTRJAeTQB1BEFUE8WwBhhMAI1hQBCkwIIDzEAkQMhshFBMAlqSwKElnIAZdEECE0UQJ2oghdfHaeVxUFtGfASqhSAsWIAIgeCoCYiCO2g5/LQ3guSmaIJYLwoeFFDGLyYXRy0koCWqCEFEJoYgLmiBiByIAggghAQekoCmCuaAMaLghe10hD+UNIxtycVenX2gWIWKQDzRQ1A5EAQvIggBIiVJQFCjyaAeqIA6soCaCcMgWA2AWiJWkLgJVzFAEQOBKEpIghBEA6SAKFFEyC0QkYvakUVvJhJGIxGbSEAdaQA1BEDsHbUAEQOBGsgghAkRkgCqFeXAOaIJnhRM6rQlLY6yXCUBi1kwItaUgDmFQMQORAEX4ggKMD7D1yiCc0v/GaVBS++OlMriIOWItASM4oBiBwIgt6IIISA1VMOoJ4ogHpRBdBeFpoSSOerpUTo2fm3hZpC4EXEoDUiB4LVEEEIkXBIOYA6aQcvaooCtO7AtBaGtjC6E1cbs0sBaCMGIHIgCIEigqACeksCmD+aAOqmH5piBmGwGloIgRcRA/+IHAhWRQRBJYySBNAmmgDqiQKoH1VoighDa7QUAlBfCkA7MQCRA0FQggiCihghCaBNNAG0FQXQRhag7c4xnKVBaxloihZiABI1EAQzIoKgMkZKAqgfTQBtRAH0kQUvvjpRq4iDnhLQEqtJgReRA0EIDREEDTBiGKQXraIJoJ0ogL6y0JRgOl4tZMLIjt8fWkkBhK8YgMiBEF6IIGhIOEYTQFtRAONkoT3M2pmrgZZC4EVrMQCRA0FQExEEjTFKEkA/UQB9ZAHMJQxWRg8h8CJiIAjWRARBB4yUBNBeFED7qIIXEQZl6CkEoI8UgLFiACIHQngjgqATRtYleNGyPsGLHlGFprTV8Yk06C8EXiJFDEDkQAh/RBB0JhKiCV70lgUvvjrHcBQHo0SgKXpJAYgYCIKeiCAYgNGSAPqKAuiXgvCHv87UzPJgBgloiZ5SAOYQAxA5ECILEQSDMEPKAYwTBTBWFlqipBMOVirM2NEHg95SAOYRAxA5ECIPEQSDMUM0AfQXBWjd4ZhJGAIh2A6/rq6eL2f8zjcz57Hynw3k5xVjt9vp2qMj4/cfzukXTKJ33xyNWqsMI6QARAwEwQzYPB6PoitAaWkpKSkpFBbPJTk5Ue12RSRmEAUveopCW1hNFtpj+eL/uGHqU6xZuZnoaCdDRvQhOyeTstIKFv+1mtLiCpxOB8++eSOTjtnXsHYaJQRezCQGIHIghCelpeV0SB1PSUkJycnJPreTCIKJMEs0AZpfqI2QBatHF5qyeMEqzj7+LsrLqjjjgklcffNpZHZMa3y9rq6ej9/+kYfumo4zSt9/kkYLgRcRA0EwHyIIJsMstQlNMSL90BKrCkP+rmIuPOU+ysuquGnaOUy9dkqrbaKinJx+/iTG7DOItPQkTdtjFiHwYjYxAJEDQfAigmBSRBT801ZHZ0ZpmHbjKxQWlHLQ4WPalIOm9BvYTdVjm00GvJhRCkDEQBBaIoJgcsyUdvBidPrBF746RKPEYd3arXz12RycTgfTHrtEs+OYVQRaImIgCNZCBMECmDGa4MWsstCU9jpQrQTiw+k/4PF4OPiIseR07xjSvqwiAS0xqxR4ETkQBN+IIFgIM4sCmCsFEQxKOt9ApOL3nxYDcOjkcUEda/Gitbz8/JcsnL+KDet3cN1NJ3HbtDODbqNRmF0KQMRAEAJBBMGCmDHt0BQrRBVCpb2O3uVysWbFZgD23qt3UBKy4I9VLFqwhnH7DqKgoCykduqFFaQARAwEIRhEECyK2aMJXiJBFtqisKAMt9sNQMdOae1s3ZyLLjuSS644GoAR/S5SvW1qYRUpABEDQVCCCILFsYooQOsOJZyFwWZTXtdgt9tVbIm6WEkKQMRAEELBvFciIShiHamWuxjGOFKbPcKJDulJJCbFAfDfmm0Gt0Y5Vv2NrPjvQRDMhghCmGHlC6MVOyJf2O12DjxkBADvvPmj323r611sWLddh1a1j1WFwIuVz39BMBsiCGGK1S+ULTsqK3ZW1910MlFRTt6d/iPPPvEpLper2et1dfXMnDGXA/e+lpW7Cxr1xurfsRern++CYEakBiHMsVKNQnu01YGZuY5h2IhevPDa1Vx58bPcfet0Xnx2FqPH9iMuNoa8vGL+/msNFeXVAAwf0Uvz9lhZAHwhUiAI2iGCECE0vZCGgyx48dXpmUUcTjh5P0aM7sv/nvuCX39ays8/LKa+3k1ah0RGjOzDhP2HctxJE+jSNVO1Y4ajCLRExEAQtEcEIQIJp6iCL/x1knrLQ6/enXn4yYtV3WckSEBbiBgIgn6IIEQwkSAKbRFI52pUBCLGkcquXcXM+W0JAFVVtaxbm8fXny8hISGWwybtbUi7jESkQBCMwebxeBRN8l5aWkpKSgqFxXNJTk5Uu12CAUSaKJiV335dzORDrmn1fLfunVjx34f6N8ggRAwEQRtKS8vpkDqekpISkpOTfW4nEQShkXCtU7Aa++0/kvK6X41uhiGIFAiCeRBBENpEZEHQExEDQTAfIghCu4gsCFogUiAI5kYEQQiKSC1sFNRBpEAQrIMIgqCIlhd6EQbBFyIFgmBNRBAEVZA0hNAUkQJBsD4iCILqiCxEHiIEghB+iCAImiKpiPBFpEAQwhsRBEFXRBisiwiBIEQWIgiCoYgwmBORAUEQRBAEU9FWxyTSoC0iA4IgtIUIgmB6RBrUQ2RAEIRAEUEQLImvjk7EoQERAUEQQkUEQQgrAukYw0EiRAAEQdAaEQQh4lDauWohFtLRC4JgVkQQBCFApDMXBCGSsBvdAEEQBEEQzIcIgiAIgiAIrRBBEARBEAShFSIIgiAIgiC0QgRBEARBEIRWiCAIgiAIgtAKEQRBEARBEFohgiAIgiAIQitEEARBEARBaIUIgiAIgiAIrRBBEARBEAShFSIIgiAIgiC0QgRBEARBEIRWiCAIgiAIgtAKEQRBEARBEFohgiAIgiAIQitEEARBEARBaIVT6Rs9Hg8ApaUVqjVGEARBEARt8fbb3n7cF4oFoaysDIAe3Q5TugtBEARBEAyirKyMlJQUn6/bPO0phA/cbje5ubkkJSVhs9kUN1AQBEEQBP3weDyUlZWRnZ2N3e670kCxIAiCIAiCEL5IkaIgCIIgCK0QQRAEQRAEoRUiCIIgCIIgtEIEQRAEQRCEVoggCIIgCILQChEEQRAEQRBaIYIgCIIgCEIrRBAEQRAEQWiFCIIgCIIgCK0QQRAEQRAEoRUiCIIgCIIgtEIEQRAEQRCEVvw/VQI9LEz8tr0AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Create a 2D grid of points\n",
    "x = np.linspace(-3.0, 3.0, 100)\n",
    "y = np.linspace(-3.0, 3.0, 100)\n",
    "X, Y = np.meshgrid(x, y)\n",
    "\n",
    "# Define a function to simulate the adversarial loss landscape\n",
    "Z = np.exp(-(X**2 + Y**2)) + 0.5 * np.exp(-((X - 2)**2 + (Y - 2)**2))\n",
    "\n",
    "# Plot the contour\n",
    "plt.contourf(X, Y, Z, levels=20, cmap='YlGnBu')\n",
    "\n",
    "# # Define the paths for FGSM-RS and PGD\n",
    "# # FGSM-RS path\n",
    "# fgsm_rs_x = [-1, 1]\n",
    "# fgsm_rs_y = [1, 0]\n",
    "# plt.plot(fgsm_rs_x, fgsm_rs_y, 'ko-', label='FGSM-RS', lw=2)\n",
    "\n",
    "# PGD path\n",
    "pgd_x = [0.25, 0.5, 0.75, 1]\n",
    "pgd_y = [0.75, 1, 1.25, 1.43]\n",
    "plt.plot(pgd_x, pgd_y, 'ro-', label='PGD', lw=2)\n",
    "\n",
    "# Mark key points with labels\n",
    "plt.text(0, 0.75, r'$x^{0}$', fontsize=15)\n",
    "# plt.text(1, 0, r'$x^{b}$', fontsize=12)\n",
    "plt.text(1, 1.6, r'$x^{adv}$', fontsize=15)\n",
    "\n",
    "# Display attack budget epsilon\n",
    "plt.text(1.1, 0, r'$\\epsilon$', fontsize=15)\n",
    "\n",
    "# Display two classes\n",
    "plt.text(0, -1.5, r'$C_1$', fontsize=15)\n",
    "plt.text(2.5, 1, r'$C_2$', fontsize=15)\n",
    "\n",
    "# add a decision boundary\n",
    "plt.plot([-2, 4.25], [4, -1.75], 'k--', label='Decision Boundary', color='black')\n",
    "plt.text(-1, 2.5, r'$f(x)$', fontsize=15)\n",
    "\n",
    "# Plot a circle around point x as the attack budget epsilon\n",
    "attack_budget_radius = 1  # Set the radius of the circle (epsilon)\n",
    "circle = plt.Circle((0.25, 0.75), attack_budget_radius, color='gray', fill=False, linestyle='--', linewidth=1.5)\n",
    "plt.gca().add_artist(circle)\n",
    "\n",
    "# Remove axis labels, ticks, and borders\n",
    "plt.gca().set_xticks([])  # Remove x-axis ticks\n",
    "plt.gca().set_yticks([])  # Remove y-axis ticks\n",
    "# plt.gca().spines['top'].set_visible(False)\n",
    "# plt.gca().spines['right'].set_visible(False)\n",
    "# plt.gca().spines['left'].set_visible(False)\n",
    "# plt.gca().spines['bottom'].set_visible(False)\n",
    "\n",
    "# Remove legend and title to maintain simplicity\n",
    "plt.legend().set_visible(False)\n",
    "\n",
    "# # set x aixs limit\n",
    "plt.xlim(-2, 3)\n",
    "plt.ylim(-2, 3)\n",
    "\n",
    "\n",
    "plt.savefig(\"Visualisation/bounded_attack.png\", dpi=300, bbox_inches = 'tight')\n",
    "# Show the plot\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\n10599070\\AppData\\Local\\Temp\\ipykernel_39976\\3938519537.py:38: UserWarning: color is redundantly defined by the 'color' keyword argument and the fmt string \"k--\" (-> color='k'). The keyword argument will take precedence.\n",
      "  plt.plot([-2, 4.25], [4, -1.75], 'k--', label='Decision Boundary', color='black')\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Create a 2D grid of points\n",
    "x = np.linspace(-3.0, 3.0, 100)\n",
    "y = np.linspace(-3.0, 3.0, 100)\n",
    "X, Y = np.meshgrid(x, y)\n",
    "\n",
    "# Define a function to simulate the adversarial loss landscape\n",
    "Z = np.exp(-(X**2 + Y**2)) + 0.5 * np.exp(-((X - 2)**2 + (Y - 2)**2))\n",
    "\n",
    "# Plot the contour\n",
    "plt.contourf(X, Y, Z, levels=20, cmap='YlGnBu')\n",
    "\n",
    "# # Define the paths for FGSM-RS and PGD\n",
    "# FGSM-RS path\n",
    "fgsm_rs_x = [0.25, 0.875]\n",
    "fgsm_rs_y = [0.75, 1.38]\n",
    "plt.plot(fgsm_rs_x, fgsm_rs_y, 'ro-', label='FGSM-RS', lw=2)\n",
    "\n",
    "# PGD path\n",
    "# pgd_x = [0.25, 0.5, 0.75, 1]\n",
    "# pgd_y = [0.75, 1, 1.25, 1.43]\n",
    "# plt.plot(pgd_x, pgd_y, 'ro-', label='PGD', lw=2)\n",
    "\n",
    "# Mark key points with labels\n",
    "plt.text(0, 0.75, r'$x^{0}$', fontsize=15)\n",
    "# plt.text(1, 0, r'$x^{b}$', fontsize=12)\n",
    "plt.text(1, 1.5, r'$x^{adv}$', fontsize=15)\n",
    "\n",
    "\n",
    "\n",
    "# Display two classes\n",
    "plt.text(0, -1.5, r'$C_1$', fontsize=15)\n",
    "plt.text(2.5, 1, r'$C_2$', fontsize=15)\n",
    "\n",
    "# add a decision boundary\n",
    "plt.plot([-2, 4.25], [4, -1.75], 'k--', label='Decision Boundary', color='black')\n",
    "plt.text(-1, 2.5, r'$f(x)$', fontsize=15)\n",
    "\n",
    "# # Display attack budget epsilon\n",
    "# plt.text(1.1, 0, r'$\\epsilon$', fontsize=15)\n",
    "\n",
    "# # Plot a circle around point x as the attack budget epsilon\n",
    "# attack_budget_radius = 1  # Set the radius of the circle (epsilon)\n",
    "# circle = plt.Circle((0.25, 0.75), attack_budget_radius, color='gray', fill=False, linestyle='--', linewidth=1.5)\n",
    "# plt.gca().add_artist(circle)\n",
    "\n",
    "# Remove axis labels, ticks, and borders\n",
    "plt.gca().set_xticks([])  # Remove x-axis ticks\n",
    "plt.gca().set_yticks([])  # Remove y-axis ticks\n",
    "# plt.gca().spines['top'].set_visible(False)\n",
    "# plt.gca().spines['right'].set_visible(False)\n",
    "# plt.gca().spines['left'].set_visible(False)\n",
    "# plt.gca().spines['bottom'].set_visible(False)\n",
    "\n",
    "# Remove legend and title to maintain simplicity\n",
    "plt.legend().set_visible(False)\n",
    "\n",
    "# # set x aixs limit\n",
    "plt.xlim(-2, 3)\n",
    "plt.ylim(-2, 3)\n",
    "\n",
    "plt.savefig(\"Visualisation/unbounded_attack.png\", dpi=300, bbox_inches = 'tight')\n",
    "# Show the plot\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Attack success rate for different datasets and models\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# Creating the dataset based on the extracted information\n",
    "data = {\n",
    "    'Dataset': ['Adult', 'Adult', 'Adult', 'German', 'German', 'German',\n",
    "                'COMPAS', 'COMPAS', 'COMPAS', 'Diabetes', 'Diabetes', 'Diabetes',\n",
    "                'Breast Cancer', 'Breast Cancer', 'Breast Cancer'],\n",
    "    'Model': ['LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC'],\n",
    "    'FGSM': [84.95, 84.65, 84.75, 80.73, 76.56, 80.73,\n",
    "             79.26, 80.82, 79.76, 75.0, 72.66, 75.78,\n",
    "             96.88, 96.88, 98.44],\n",
    "    'PGD': [84.95, 85.15, 84.75, 80.73, 78.65, 80.73,\n",
    "            79.26, 80.89, 79.76, 75.0, 72.66, 75.78,\n",
    "            96.88, 96.88, 98.44],\n",
    "    'C&W': [69.86, 45.19, 14.76, 72.4, 61.98, 18.23,\n",
    "            69.46, 72.09, 20.81, 75.0, 71.88, 25.78,\n",
    "            90.63, 82.81, 4.69],\n",
    "    'DeepFool': [82.69, 84.51, 85.32, 81.25, 80.73, 81.25,\n",
    "                 79.33, 80.47, 79.76, 78.13, 72.66, 75.78,\n",
    "                 98.44, 96.88, 98.44],\n",
    "    'LowProFool': [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan,\n",
    "                   np.nan, np.nan, np.nan, 75.78, 72.66, 75.78,\n",
    "                   98.44, 96.88, 98.44]\n",
    "}\n",
    "\n",
    "# Convert to DataFrame\n",
    "df = pd.DataFrame(data)\n",
    "\n",
    "# Create a combined column for Dataset and Model\n",
    "df['Dataset_Model'] = df['Dataset'] + ' - ' + df['Model']\n",
    "\n",
    "pivot_df = df[['Dataset_Model', 'FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']]\n",
    "\n",
    "# plot the heatmap\n",
    "sns.set_theme(style='white')\n",
    "plt.figure(figsize=(7, 8))\n",
    "ax = sns.heatmap(pivot_df.set_index('Dataset_Model'), annot=True, linewidth=.5, cmap='coolwarm', fmt=\".2f\", vmin=0, vmax=100)\n",
    "ax.set(xlabel=\"\", ylabel=\"\")\n",
    "# plt.title('Accuracy of Different Attacks on Different Models and Datasets')\n",
    "# plt.show()\n",
    "\n",
    "plt.savefig(\"Visualisation/asr_rgb.png\", dpi=300, bbox_inches = 'tight')  # Save as PNG in RGB\n",
    "plt.savefig(\"Visualisation/asr_rgb.pdf\", format='pdf', dpi=300, bbox_inches = 'tight')  # Save as PDF in RGB\n",
    "plt.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plot saved in CMYK mode as 'asr_cmyk.pdf'.\n"
     ]
    }
   ],
   "source": [
    "from PIL import Image\n",
    "\n",
    "# Step 2: Convert RGB to CMYK and save as PDF\n",
    "rgb_image = Image.open(\"Visualisation/asr_rgb.png\")\n",
    "cmyk_image = rgb_image.convert(\"CMYK\")\n",
    "\n",
    "# Save the CMYK image as a PDF\n",
    "cmyk_image.save(\"Visualisation/asr_cmyk.pdf\", \"PDF\")\n",
    "\n",
    "print(\"Plot saved in CMYK mode as 'asr_cmyk.pdf'.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plot saved in CMYK mode as 'l2_cmyk.pdf'.\n"
     ]
    }
   ],
   "source": [
    "# Attack success rate for different datasets and models\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# Extracted data\n",
    "data = {\n",
    "    'Dataset': ['Adult', 'Adult', 'Adult', 'German', 'German', 'German',\n",
    "                'COMPAS', 'COMPAS', 'COMPAS', 'Diabetes', 'Diabetes', 'Diabetes',\n",
    "                'Breast Cancer', 'Breast Cancer', 'Breast Cancer'],\n",
    "    'Model': ['LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC'],\n",
    "    'FGSM': [0.5644, 0.5662, 0.5656, 0.5837, 0.5932, 0.5840,\n",
    "             0.5289, 0.5187, 0.5314, 0.7842, 0.7939, 0.7808,\n",
    "             1.5263, 1.4229, 1.4978],\n",
    "    'PGD': [0.5644, 0.5663, 0.5656, 0.5837, 0.5826, 0.5840,\n",
    "            0.5289, 0.5109, 0.5314, 0.7842, 0.7999, 0.7808,\n",
    "            1.5263, 1.4672, 1.4978],\n",
    "    'C&W': [0.5217, 0.1866, 0.0006, 0.6234, 0.4396, 0.0001,\n",
    "            0.2426, 0.2459, 0.0017, 0.1942, 0.2114, 0.0091,\n",
    "            0.4392, 0.2793, 0.0018],\n",
    "    'DeepFool': [0.6355, 0.9627, 0.1108, 0.6435, 1.1269, 0.4128,\n",
    "                 0.4108, 0.4637, 0.2360, 0.2719, 0.2912, 0.2305,\n",
    "                 1.7005, 1.5350, 0.3923],\n",
    "    'LowProFool': [None, None, None, None, None, None,\n",
    "                   None, None, None, 0.7545, 0.6017, 0.6557,\n",
    "                   1.3016, 1.7114, 0.9481]\n",
    "}\n",
    "\n",
    "# Convert to DataFrame\n",
    "df = pd.DataFrame(data)\n",
    "\n",
    "# Create a combined column for Dataset and Model\n",
    "df['Dataset_Model'] = df['Dataset'] + ' - ' + df['Model']\n",
    "\n",
    "pivot_df = df[['Dataset_Model', 'FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']]\n",
    "\n",
    "# plot the heatmap\n",
    "sns.set_theme(style='white')\n",
    "plt.figure(figsize=(7, 8))\n",
    "ax = sns.heatmap(pivot_df.set_index('Dataset_Model'), annot=True, linewidth=.5, cmap='coolwarm', fmt=\".2f\")\n",
    "ax.set(xlabel=\"\", ylabel=\"\")\n",
    "# plt.title('Accuracy of Different Attacks on Different Models and Datasets')\n",
    "# plt.show()\n",
    "\n",
    "plt.savefig(\"Visualisation/l2_rgb.png\", dpi=300, bbox_inches = 'tight')  # Save as PNG in RGB\n",
    "plt.savefig(\"Visualisation/l2_rgb.pdf\", format='pdf', dpi=300, bbox_inches = 'tight')  # Save as PDF in RGB\n",
    "plt.close()\n",
    "\n",
    "\n",
    "# Step 2: Convert RGB to CMYK and save as PDF\n",
    "rgb_image = Image.open(\"Visualisation/l2_rgb.png\")\n",
    "cmyk_image = rgb_image.convert(\"CMYK\")\n",
    "\n",
    "# Save the CMYK image as a PDF\n",
    "cmyk_image.save(\"Visualisation/l2_cmyk.pdf\", \"PDF\")\n",
    "\n",
    "print(\"Plot saved in CMYK mode as 'l2_cmyk.pdf'.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plot saved in CMYK mode as 'linf_cmyk.pdf'.\n"
     ]
    }
   ],
   "source": [
    "# Linf for different datasets and models\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# Extracted data\n",
    "\n",
    "data = {\n",
    "    'Dataset': ['Adult', 'Adult', 'Adult', 'German', 'German', 'German',\n",
    "                'COMPAS', 'COMPAS', 'COMPAS', 'Diabetes', 'Diabetes', 'Diabetes',\n",
    "                'Breast Cancer', 'Breast Cancer', 'Breast Cancer'],\n",
    "    'Model': ['LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC'],\n",
    "    'FGSM': [0.3000, 0.2999, 0.3000, 0.3000, 0.3000, 0.3000,\n",
    "             0.3000, 0.3000, 0.3000, 0.3000, 0.3000, 0.3000,\n",
    "             0.3000, 0.3000, 0.3000],\n",
    "    'PGD': [0.3000, 0.3000, 0.3000, 0.3000, 0.3000, 0.3000,\n",
    "            0.3000, 0.3000, 0.3000, 0.3000, 0.3000, 0.3000,\n",
    "            0.3000, 0.3000, 0.3000],\n",
    "    'C&W': [0.4081, 0.1570, 0.0005, 0.4146, 0.3569, 0.0001,\n",
    "            0.2147, 0.2137, 0.0015, 0.1491, 0.1393, 0.0066,\n",
    "            0.1854, 0.1502, 0.0009],\n",
    "    'DeepFool': [0.4189, 0.4154, 0.1086, 0.3364, 0.4170, 0.3027,\n",
    "                 0.3257, 0.3198, 0.1833, 0.1938, 0.1701, 0.1592,\n",
    "                 0.5240, 0.5677, 0.1408],\n",
    "    'LowProFool': [None, None, None, None, None, None,\n",
    "                   None, None, None, 0.5228, 0.3471, 0.4548,\n",
    "                   0.4537, 0.7110, 0.3485]\n",
    "}\n",
    "\n",
    "# Convert to DataFrame\n",
    "df = pd.DataFrame(data)\n",
    "\n",
    "# Create a combined column for Dataset and Model\n",
    "df['Dataset_Model'] = df['Dataset'] + ' - ' + df['Model']\n",
    "\n",
    "pivot_df = df[['Dataset_Model', 'FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']]\n",
    "\n",
    "# plot the heatmap\n",
    "sns.set_theme(style='white')\n",
    "plt.figure(figsize=(7, 8))\n",
    "ax = sns.heatmap(pivot_df.set_index('Dataset_Model'), annot=True, linewidth=.5, cmap='coolwarm', fmt=\".2f\", vmin=0, vmax=1)\n",
    "ax.set(xlabel=\"\", ylabel=\"\")\n",
    "# plt.title('Accuracy of Different Attacks on Different Models and Datasets')\n",
    "# plt.show()\n",
    "\n",
    "plt.savefig(\"Visualisation/linf_rgb.png\", dpi=300, bbox_inches = 'tight')  # Save as PNG in RGB\n",
    "plt.savefig(\"Visualisation/linf_rgb.pdf\", format='pdf', dpi=300, bbox_inches = 'tight')  # Save as PDF in RGB\n",
    "plt.close()\n",
    "\n",
    "# Step 2: Convert RGB to CMYK and save as PDF\n",
    "\n",
    "rgb_image = Image.open(\"Visualisation/linf_rgb.png\")\n",
    "cmyk_image = rgb_image.convert(\"CMYK\")\n",
    "\n",
    "# Save the CMYK image as a PDF\n",
    "cmyk_image.save(\"Visualisation/linf_cmyk.pdf\", \"PDF\")\n",
    "\n",
    "print(\"Plot saved in CMYK mode as 'linf_cmyk.pdf'.\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plot saved in CMYK mode as 'sparsity_cmyk.pdf'.\n"
     ]
    }
   ],
   "source": [
    "# Sparsity for different datasets and models\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# Extracted data\n",
    "\n",
    "num_of_features = [102, 58, 23, 8, 30]\n",
    "\n",
    "data = {\n",
    "    'Dataset': ['Adult', 'Adult', 'Adult', 'German', 'German', 'German',\n",
    "                'COMPAS', 'COMPAS', 'COMPAS', 'Diabetes', 'Diabetes', 'Diabetes',\n",
    "                'Breast Cancer', 'Breast Cancer', 'Breast Cancer'],\n",
    "    'Model': ['LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC'],\n",
    "    'FGSM': [3.6779, 3.6890, 3.6946, 4.3073, 4.4219, 4.3125,\n",
    "             3.9311, 3.9197, 3.9361, 7.4375, 7.5391, 7.4609,\n",
    "             29.9688, 29.8438, 29.9375],\n",
    "    'PGD': [3.6779, 3.6920, 3.6946, 4.3073, 4.3281, 4.3125,\n",
    "            3.9311, 3.9006, 3.9361, 7.4375, 7.6250, 7.4609,\n",
    "            29.9688, 29.8594, 29.9375],\n",
    "    'C&W': [3.7146, 3.0948, 2.0699, 5.0104, 4.7344, 3.6927,\n",
    "            3.8040, 3.8111, 3.2003, 8.0000, 8.0000, 6.8438,\n",
    "            29.8125, 29.8125, 29.7031],\n",
    "    'DeepFool': [3.9497, 4.7433, 3.6946, 4.6458, 6.0000, 4.3125,\n",
    "                 3.9382, 3.9929, 3.9361, 7.4375, 7.5391, 7.4609,\n",
    "                 29.9688, 29.8438, 29.9375],\n",
    "    'LowProFool': [None, None, None, None, None, None,\n",
    "                   None, None, None, 7.4375, 7.5547, 7.4609,\n",
    "                   29.9688, 29.9531, 29.9375]\n",
    "}\n",
    "\n",
    "# Convert to DataFrame\n",
    "df = pd.DataFrame(data)\n",
    "\n",
    "# Dataset is Adult then num_of_features[0] = 102, FGSM, PGD, C&W, DeepFool, LowProFool value = original value / num_of_features[0]\n",
    "# Dataset is German then num_of_features[1] = 30, FGSM, PGD, C&W, DeepFool, LowProFool value = original value / num_of_features[1]\n",
    "# Dataset is COMPAS then num_of_features[2] = 23, FGSM, PGD, C&W, DeepFool, LowProFool value = original value / num_of_features[2]\n",
    "# Dataset is Diabetes then num_of_features[3] = 8, FGSM, PGD, C&W, DeepFool, LowProFool value = original value / num_of_features[3]\n",
    "# Dataset is Breast Cancer then num_of_features[4] = 30, FGSM, PGD, C&W, DeepFool, LowProFool value = original value / num_of_features[4]\n",
    "# Transform the data based on the number of features\n",
    "\n",
    "for attack in ['FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']:\n",
    "    df[attack] = df.apply(lambda x: x[attack] / num_of_features[0] if x['Dataset'] == 'Adult' else\n",
    "    x[attack] / num_of_features[1] if x['Dataset'] == 'German' else\n",
    "    x[attack] / num_of_features[2] if x['Dataset'] == 'COMPAS' else\n",
    "    x[attack] / num_of_features[3] if x['Dataset'] == 'Diabetes' else\n",
    "    x[attack] / num_of_features[4], axis=1)\n",
    "\n",
    "# Create a combined column for Dataset and Model\n",
    "df['Dataset_Model'] = df['Dataset'] + ' - ' + df['Model']\n",
    "\n",
    "pivot_df = df[['Dataset_Model', 'FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']]\n",
    "\n",
    "# plot the heatmap\n",
    "sns.set_theme(style='white')\n",
    "plt.figure(figsize=(7, 8))\n",
    "ax = sns.heatmap(pivot_df.set_index('Dataset_Model'), annot=True, linewidth=.5, cmap='coolwarm', fmt=\".2f\", vmin=0, vmax=1)\n",
    "ax.set(xlabel=\"\", ylabel=\"\")\n",
    "# plt.title('Accuracy of Different Attacks on Different Models and Datasets')\n",
    "# plt.show()\n",
    "\n",
    "plt.savefig(\"Visualisation/sparsity_rgb.png\", dpi=300, bbox_inches = 'tight')  # Save as PNG in RGB\n",
    "plt.savefig(\"Visualisation/sparsity_rgb.pdf\", format='pdf', dpi=300, bbox_inches = 'tight')  # Save as PDF in RGB\n",
    "plt.close()\n",
    "\n",
    "# Step 2: Convert RGB to CMYK and save as PDF\n",
    "rgb_image = Image.open(\"Visualisation/sparsity_rgb.png\")\n",
    "cmyk_image = rgb_image.convert(\"CMYK\")\n",
    "\n",
    "# Save the CMYK image as a PDF\n",
    "cmyk_image.save(\"Visualisation/sparsity_cmyk.pdf\", \"PDF\")\n",
    "\n",
    "print(\"Plot saved in CMYK mode as 'sparsity_cmyk.pdf'.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plot saved in CMYK mode as 'deviation_cmyk.pdf'.\n"
     ]
    }
   ],
   "source": [
    "# deviation for different datasets and models\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from PIL import Image\n",
    "\n",
    "# Extracted data\n",
    "\n",
    "data = {\n",
    "    'Dataset': ['Adult', 'Adult', 'Adult', 'German', 'German', 'German',\n",
    "                'COMPAS', 'COMPAS', 'COMPAS', 'Diabetes', 'Diabetes', 'Diabetes',\n",
    "                'Breast Cancer', 'Breast Cancer', 'Breast Cancer'],\n",
    "    'Model': ['LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC'],\n",
    "    'FGSM': [0.0787, 0.0779, 0.0788, 0.1519, 0.1654, 0.1521,\n",
    "             0.0576, 0.0568, 0.0577, 0.1375, 0.1463, 0.1342,\n",
    "             0.7075, 0.2807, 0.5639],\n",
    "    'PGD': [0.0787, 0.0786, 0.0788, 0.1519, 0.1595, 0.1521,\n",
    "            0.0576, 0.0563, 0.0577, 0.1375, 0.1492, 0.1342,\n",
    "            0.7075, 0.3552, 0.5639],\n",
    "    'C&W': [0.0773, 0.0277, 0.0001, 0.1540, 0.0999, 0.0000,\n",
    "            0.0270, 0.0277, 0.0002, 0.0326, 0.0344, 0.0014,\n",
    "            0.2088, 0.0623, 0.0006],\n",
    "    'DeepFool': [0.1274, 0.2861, 0.0089, 0.2109, 0.4722, 0.0859,\n",
    "                 0.0513, 0.0856, 0.0290, 0.0500, 0.0560, 0.0384,\n",
    "                 0.8427, 0.4171, 0.1540],\n",
    "    'LowProFool': [None, None, None, None, None, None,\n",
    "                   None, None, None, 0.1365, 0.1128, 0.1081,\n",
    "                   0.6694, 0.4691, 0.3788]\n",
    "}\n",
    "\n",
    "# Convert to DataFrame\n",
    "df = pd.DataFrame(data)\n",
    "\n",
    "# Create a combined column for Dataset and Model\n",
    "df['Dataset_Model'] = df['Dataset'] + ' - ' + df['Model']\n",
    "\n",
    "pivot_df = df[['Dataset_Model', 'FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']]\n",
    "\n",
    "# plot the heatmap\n",
    "sns.set_theme(style='white')\n",
    "plt.figure(figsize=(7, 8))\n",
    "ax = sns.heatmap(pivot_df.set_index('Dataset_Model'), annot=True, linewidth=.5, cmap='coolwarm', fmt=\".2f\", vmin=0, vmax=1)\n",
    "ax.set(xlabel=\"\", ylabel=\"\")\n",
    "# plt.title('Accuracy of Different Attacks on Different Models and Datasets')\n",
    "# plt.show()\n",
    "\n",
    "plt.savefig(\"Visualisation/deviation_rgb.png\", dpi=300, bbox_inches = 'tight')  # Save as PNG in RGB\n",
    "plt.savefig(\"Visualisation/deviation_rgb.pdf\", format='pdf', dpi=300, bbox_inches = 'tight')  # Save as PDF in RGB\n",
    "plt.close()\n",
    "\n",
    "# Step 2: Convert RGB to CMYK and save as PDF\n",
    "\n",
    "rgb_image = Image.open(\"Visualisation/deviation_rgb.png\")\n",
    "cmyk_image = rgb_image.convert(\"CMYK\")\n",
    "\n",
    "# Save the CMYK image as a PDF\n",
    "cmyk_image.save(\"Visualisation/deviation_cmyk.pdf\", \"PDF\")\n",
    "\n",
    "print(\"Plot saved in CMYK mode as 'deviation_cmyk.pdf'.\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plot saved in CMYK mode as 'sensitivity_cmyk.pdf'.\n"
     ]
    }
   ],
   "source": [
    "# sensitivity for different datasets and models\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from PIL import Image\n",
    "\n",
    "# Extracted data\n",
    "\n",
    "data = {\n",
    "    'Dataset': ['Adult', 'Adult', 'Adult', 'German', 'German', 'German',\n",
    "                'COMPAS', 'COMPAS', 'COMPAS', 'Diabetes', 'Diabetes', 'Diabetes',\n",
    "                'Breast Cancer', 'Breast Cancer', 'Breast Cancer'],\n",
    "    'Model': ['LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC', 'LR', 'MLP', 'LinearSVC',\n",
    "              'LR', 'MLP', 'LinearSVC'],\n",
    "    'FGSM': [2.4898, 2.4998, 2.5020, 1.1172, 1.1397, 1.1178,\n",
    "             2.8215, 2.6871, 2.8485, 1.6910, 1.7464, 1.6867,\n",
    "             1.9884, 1.8176, 1.9371],\n",
    "    'PGD': [2.4898, 2.5019, 2.5020, 1.1172, 1.1130, 1.1178,\n",
    "            2.8215, 2.5944, 2.8485, 1.6910, 1.7691, 1.6867,\n",
    "            1.9884, 1.9000, 1.9371],\n",
    "    'C&W': [1.5780, 0.4857, 0.0016, 1.0668, 0.6896, 0.0002,\n",
    "            0.7602, 0.7803, 0.0047, 0.2931, 0.3613, 0.0147,\n",
    "            0.3773, 0.2367, 0.0016],\n",
    "    'DeepFool': [2.3923, 2.6746, 0.4252, 0.9765, 1.0908, 0.8187,\n",
    "                 1.6436, 1.8902, 0.9090, 0.4780, 0.5704, 0.4114,\n",
    "                 1.8599, 1.6574, 0.4496],\n",
    "    'LowProFool': [None, None, None, None, None, None,\n",
    "                   None, None, None, 1.3519, 1.1873, 1.1751,\n",
    "                   1.3337, 1.8253, 1.0620]\n",
    "}\n",
    "\n",
    "\n",
    "# Convert to DataFrame\n",
    "df = pd.DataFrame(data)\n",
    "\n",
    "# Create a combined column for Dataset and Model\n",
    "df['Dataset_Model'] = df['Dataset'] + ' - ' + df['Model']\n",
    "\n",
    "pivot_df = df[['Dataset_Model', 'FGSM', 'PGD', 'C&W', 'DeepFool', 'LowProFool']]\n",
    "\n",
    "# plot the heatmap\n",
    "sns.set_theme(style='white')\n",
    "plt.figure(figsize=(7, 8))\n",
    "ax = sns.heatmap(pivot_df.set_index('Dataset_Model'), annot=True, linewidth=.5, cmap='coolwarm', fmt=\".2f\", vmin=0, vmax=1)\n",
    "ax.set(xlabel=\"\", ylabel=\"\")\n",
    "# plt.title('Accuracy of Different Attacks on Different Models and Datasets')\n",
    "# plt.show()\n",
    "\n",
    "plt.savefig(\"Visualisation/sensitivity_rgb.png\", dpi=300, bbox_inches = 'tight')  # Save as PNG in RGB\n",
    "plt.savefig(\"Visualisation/sensitivity_rgb.pdf\", format='pdf', dpi=300, bbox_inches = 'tight')  # Save as PDF in RGB\n",
    "plt.close()\n",
    "\n",
    "# Step 2: Convert RGB to CMYK and save as PDF\n",
    "\n",
    "rgb_image = Image.open(\"Visualisation/sensitivity_rgb.png\")\n",
    "cmyk_image = rgb_image.convert(\"CMYK\")\n",
    "\n",
    "# Save the CMYK image as a PDF\n",
    "cmyk_image.save(\"Visualisation/sensitivity_cmyk.pdf\", \"PDF\")\n",
    "\n",
    "print(\"Plot saved in CMYK mode as 'sensitivity_cmyk.pdf'.\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "xai",
   "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": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}