Kohonen_(KSOFM).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Kohonen (KSOFM).ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0oCI7ZGmf7yW"
      },
      "source": [
        "#Kohonen's Self-Organizing Feature Map (KSOFM)\n",
        "\n",
        "We want to viualize 3D input (rgb) into 2D using SOM network. Random selection of original image pixels are given to the SOM algorithm, which then returns another photo with organized pixel's color"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zhzy9gxMaC-W"
      },
      "source": [
        "import math\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1xSbHPdtsyD5"
      },
      "source": [
        "class SOM :\n",
        "    def __init__ (self, input_data, input_shape=(40, 40, 3), sigma=4) :\n",
        "        \n",
        "        self.input_data = input_data\n",
        "\n",
        "        self.weight = self.initial_weight(input_shape)\n",
        "        \n",
        "        self.map_index = [(i, j) for i in range(input_shape[0]) \\\n",
        "                          for j in range(input_shape[1])]\n",
        "        \n",
        "        self.sigma = sigma\n",
        "    \n",
        "    #step 1 : initialization \n",
        "    def initial_weight(self, input_shape) :\n",
        "        return np.array([np.random.choice(np.arange(0, 256), input_shape[2])\\\n",
        "                         for x in range (input_shape[0]*input_shape[1])])\\\n",
        "                         .reshape(input_shape).astype(float)\n",
        "\n",
        "    #step 2 : competition \n",
        "    def calculate_nearest_node(self, input_):\n",
        "        min_value, min_value_index = np.linalg.norm(input_ - self.weight[0][0] ), (0, 0)\n",
        "        for j in self.map_index :\n",
        "            _min_value = np.linalg.norm(input_ - self.weight[j])\n",
        "            if min_value > _min_value :\n",
        "                min_value = _min_value\n",
        "                min_value_index = j\n",
        "        return  min_value_index     \n",
        "\n",
        "\n",
        "    #step 3 : Cooperation \n",
        "    def neighborhood_function (self, dist) :\n",
        "        return np.exp(-(dist**2) / (2*self.sigma**2))\n",
        "    \n",
        "    # step 4 : Adaptation\n",
        "    def update_weight(self, learning_rate, h, x):\n",
        "        return learning_rate*h*x\n",
        "\n",
        "    def train(self, epoch, learning_rate) :\n",
        "        for e in range(epoch):\n",
        "            print(e)\n",
        "            for i in self.map_index : \n",
        "                winner = self.calculate_nearest_node(self.input_data[i]) \n",
        "                for k in  self.map_index :\n",
        "                    dist = np.sum(np.power(np.array(winner) - np.array(k), 2)) ** 0.5\n",
        "                    h = self.neighborhood_function(dist)    \n",
        "                    self.weight[k] += self.update_weight(learning_rate, h, self.input_data[i] - self.weight[k])\n",
        "           \n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GSHh8qFRij1D",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 267
        },
        "outputId": "ab8b014c-7f99-40b6-fd7b-1a8a28b69de8"
      },
      "source": [
        "def generate_random_color(dim = 1600, input_shape=(40, 40, 3)):\n",
        "    color = np.array([np.random.choice(np.arange(0, 256), 3) for x in range (dim)])\n",
        "    color = color.reshape(input_shape)\n",
        "    plt.imshow(color)\n",
        "    return color\n",
        "\n",
        "data_set = generate_random_color()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD6CAYAAABnLjEDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO2dd3yO5/fHP5eInYoVm6BK1WxR1K5NqV2rVu1Rs0ZrxNZabY2WSu1dW80aVdQeRRAjCLFiiyBcvz+SfH95+jn6zVdIcZ/36+XF8/G5nud+xsmd59znOsdYa6EoyutPnH/7ABRFiR002BXFIWiwK4pD0GBXFIegwa4oDkGDXVEcQoyC3RhTyRhz3Bhz0hjT+3kdlKIozx/zrNfZjTFuAE4AKA8gEMBuAA2stUeftiZunBQ2XpyMrveT4aroTeuRgLTz594QvW+E3iIt1DtU9D48F480Ty9P0Rsn6CJpCeKy7+L9G+L6HCY+aSfc3UVvWFgS0jJmvid6z55LR1r69JdFbxJPQ9q1IF4PAB6ZHrJ24jFp1+8Gi+sTvcX36xHM9wkAt5PeJy30jPw+XDFnSXsnc2bRe+DRXRbjXiHJKzStuN4zEX/uQoLkc+Jdj9OkxbmcTfTecDtMWqrsGQUn8PjmddLS3JU/+9fTBbk+zqXHuHfrCb/pAISPbrQpDOCktfY0ABhj5gGoAeCpwR4vTkZk91jnqn31k+jtVzY7aV3alxO95fzWkuY32U/0nuuYgbTaHWuI3oRDfUjLnuwJaYOOLBLXr4jHb3xFL358ALh2rQhpA4fsEr3t2g8hrWvfMaK3eHX+oPoO4vUAUHxCAGnlK/IPsjlbZojrC0wZRFrp6YGid13VQ6T5NakleickaE3ajpE/iN7kF35nMdUEkpr6fymur13gHdL2DuIfAADwR6lPSEs8Tv4sLEySk7SWvoNF761VC0nrsUX+7M8d6PoZHd/+pugDYvZrfHoA56PcDozQFEV5CXnhCTpjTGtjzB5jzJ6wJ/Kvf4qivHhiEuwXAET90pEhQnPBWjvZWlvQWlswbpwUMXg4RVFiQky+s+8GkN0YkwXhQf4JgIb/tCBdxmAMHOj6fa9/6kqit1IxTqSlr3RN9FY+lJW0gte/F73+xfh7UonmctKs+l99SGv3c1XShrQpI67//Mx20qaVl79TbZv7EWl7J8vfrSc04eeWvm4P0Tso/ljSxl6tLXoT9m9F2uExs0l7q+kwcX3vRtVIGzD+tugdfNyLtH6ZD4jeLtdPkjbrhw9Fb+GWaUhrs24NaYuXLRHX502cm7RrVd1Eb/BbpUi7cFhOWfWcHEJag4UPRG+Bw5zPSNdxuegdPMf1sxAaPFL0ATEIdmttmDGmI4C1ANwA+Fprjzzr/SmK8mKJyZkd1tpfAfz6nI5FUZQXiFbQKYpD0GBXFIegwa4oDiFG39n/V0LP3cXJzttctGWXvhO9jxe2Ia3qEc5+AkCum5zpbD3qM9l7rTGLw6aJ3ifr+IpApX25SLuwu5+4/oth50n7qSBX5QHAMj/2Ftkvl5o+MMlZa5xJ9B5pyNVjDRfL2xhue44gLSA+vz/Bs98V168uNpG0LFNKiN5kb7Ql7eIlzqQDQP6Oj0ircvc30dv/OJd/+ywrStrA/lPF9X0T1SOtu00teot246rFrL04kw4AH1u+ojD3K/mKwJrDDfi4Gsm574RbXN+fh7u4NDgSPbMrikPQYFcUh6DBrigOQYNdURxCrCboUuVJhzYbXRNUqbctE70dfuH9v21zVRS9f66eSdrGFnKyp1gm3mq4KF0W0bsrdBRpK/q8TdrnG+Qtn/Wr/UVaQLHhonfCQy6dPL6ioOj9ZCOX1obslfsCXKp7ibSm4zmRBgCjrnQn7e62vKTtLC5vt7x95GvSlpTnZBMANJ+Uh7Q3hvDeeQCYeIGTtT3LlRS9yeYXIu39I+1ICyw3V1xfpS/vMZ8/4hfR+/5YTj7a/vy8AOArH06wdZ/PyUAAqLGCS2M93faL3uxw3UZ9GPx+R6JndkVxCBrsiuIQNNgVxSFosCuKQ9BgVxSHEKvZeL9r91F4ims54c5Ecnfb9/YnJK3SQaGZIAD/q9xw4PbC46K3XjLuwlo08xnRWyIPZzZLFK1L2qA23HgQAN6Pw1n+C8vl0tpPn5wi7c9vq4je1QW4jPfU1HWCE/h5HJeltr8vN0L4sStnvZd8U560klUDxPW1x/FViUlPqovenn24icd2X7kcesNebi4S1kN+bW6Mm0Xa0OJ8laBDTW54AgBJmvL747OHX0MAqCWUtSbJ7iF68yRKRtqN2nzlAABa9T5HWput3FgEAAKWunZWHnpTztoDemZXFMegwa4oDkGDXVEcQoy+sxtjAgDcAfAYQJi1Vi75UhTlX+d5JOjKWGvltq9/wyOuQckUrsmWkV5y2WK63h+T1vfCQNGbf+Ne0j7IJydg2q7jBEy1i6lEb/15nOw5X2MAafmaymWPhXdUIG3LEe7ACgBzP+GSzmT+cmJoZldOKAaslPcxT27GJbtxR8tjlja58TH0asadaB90aiauD2zNk1PO7edpPQBw6gR35H3g9bPonbMniLQfq8lTfLxnFSAtwceckEyYp7+4vvgSfn9+7DZJ9OatVZi0uEe3CU5gXSsuRZ6aSS7pznJ6HGnFsnA3XgAoenyBy+0fZsjdfAH9NV5RHENMg90CWGeM2WuM4YFciqK8NMT01/ji1toLxhgvAOuNMcestS4XwyN+CLQGgCTJU8bw4RRFeVZidGa31l6I+PsKgCUIn+z6d89/xj8l8JDHziqK8uJ55mA3xiQ2xnhE/htABQC8CV1RlJeCmPwanxrAEmNM5P3MsdbyUK0o3Dvrjt3t0rlo31aKL3q7/fwWads7yPf7RTHuinouwR+it0o1LkX8epPcibbmpXmkTVm1irS9veQ5a0PbjidtRq3Nojfpas6810wqD8LMuob1GaOSit4F5/n53vtlt+hNvXEFab6nuUnEnkeTxfWTb9FcTzQu7yt6L1Tn51uupfybX/8pfGXl2uYpoveT3dxlt5o7X6157MHvLQD8tpBLpD/f3E30fvYXh8/AU+1Fb91f+bnNaiCXTq8+xVdxkqe8J3qvNB7ocjvoTBfRB8Rs1ttpAPmedb2iKLGLXnpTFIegwa4oDkGDXVEcQqzuZ0+UMhT5a/m7aM0q9xS9GT/gEskKR+RkXpakT0grkfKu6K3vkZm07l6bRG/7FVyWunnIMdIKl+O9ygDQzpfzlWPavSl6C9zjJNK8nfLYoQa+iUmbtKuX6HXvzuWTiy5xCS0ALAjmPfVvNa1F2trBs8X142Z8StqmsnJap9fCi6Q9COO98wBwNB/vcy80/6jo7bGwOWnZe3PPhHJeo8X1Zztv5sf64rroXeLDSWTfkBOit+nbG0nbunKI6A1ry0nVFZd/FL0dVrpWqs+rJ++nB/TMriiOQYNdURyCBruiOAQNdkVxCBrsiuIQjLVyd9cX8mDpUlnTyrUZws5dcjnl53FKk7Yhe6joXZ6ZM50VfigrenvN5jLLkkd/EL3lmnADDZ/GZ0mbdIMzywDQoS1n+YNPys8h1TXu7DrK5wvRez+AM+QjR/AMOgDouIqz1vfKypnsD+LlJy3XNT/S3LvKjTK+zctdYIN+HCp6RwRyCWqPSkVF78Wew0g7N0Z+zQtU5Wz8cMPlzMW/3SmuT5o6gLRk7eXuwwtOc/12u+uPRG8Jby6BHT9kvujNcpffh/or5Ss+W9IVcbndY3dBnLy9x0hePbMrikPQYFcUh6DBrigOQYNdURxCrCbocidPan8pX9xF++iyPAKnQ0ZO1nRZ1lC+40M8FqpPhj6iddC+RaQlnHpE9H6U5SfSGozkREvikjfE9SVCucxz+kc5Re9Ej29I2/qOPMLqhB/vBb92S96//2nVD0irv/G06E26+iFpGxbxCKvbRbaK6zt5cXnwD2nk5FbSD/l+8yWXO9Hmb8DjtWwF7vwLAH7bOEFXeWQ20iosl8upt9z1J+3YrTmi98l6/owu2MEltACwtyh3qP00FZduA8DRw/zatJ87SPRuP+yaaJy3aiYuB1/SBJ2iOBkNdkVxCBrsiuIQNNgVxSH81/3sxhhfANUAXLHW5o7QkgOYD8AbQACAetZaOUsVhWu3POG7ynW8Tu2fW4re7tN5b/MwN3mkU9Y0y0gb5D5c9P7UeTFpv5RsKnr7NuS959NwlbQkySeK67u+2Yi0sWNbiN76d3g/ewK7XvSe/1vTTgAo2CKD6M11nUcyba0pz7nP24ETfxfns/dwTm6kCQDNCvXm48rCjw8AJdJy8rHkW/I++SQhnNTs050bYQJAgVzTSZv0JlekhZQZJa4/doYbQ/q/LSeRb5fivgKeZ+uL3k5rN5A22v+k6I2/lmesf+3+legts/xbl9vmLo+6iiQ6Z/ZpACr9TesN4DdrbXYAv0XcVhTlJea/BnvEhJe/t+qoASDyR+h0AFxErijKS8WzfmdPba2N7Bt1CeE95EWMMa2NMXuMMXvuW/napqIoL54YJ+hseFXOUytzoo5/SmiSxPThFEV5Rp412C8bY9ICQMTf8p5HRVFeGp61u+xyAE0BjIj4m9PhAmly3sEXcze7aPkq1BW97bpyyWDDiw1Eb60JXOI448qXondj/8uk1V0RKHq/u36LtEZv7iNt6e3ipAHAr+24G2769jtE77yE75P2xQMv0dsu92PSRs6T34LdnaeRliPbO6L3l+NzSZs/hUtgR2OMuH70VM5wb+vaSfQGfXuItNK55W6rHzYuTVqm0qwBgNsmfs0ebv+TtLnF8ojrv6+cirQ8/eRs/J6kfJUgzndy9+C+R0uT1n2Q/LV22hY+B/+c5qboHbgr2OX2g1pyB2YgGmd2Y8xcADsA5DDGBBpjWiI8yMsbY/wBlIu4rSjKS8x/PbNba+XTKfDhcz4WRVFeIFpBpygOQYNdURxCrI5/so8S4uEl18TImFQDRe/AUXzpvmVneQ9zJp9xpNXPmV30JvqeEziT3ORy2aapuLS1ZDdOKI5Iy+OnAODAOU5kLYorJwPT9OO90V55qojeS79zImvuGTmZd6I6JzqLecgNNoeNX03a1EW8/z9OTU6uAcD+otzMc+0VOSG5pTPPG/ffL+8F/2MrjwJrGSaPDdu8l8t7EzXj93HpBrmc+smBGaS1+U7uK1BsMidV+1z4WvQ2PsHV5MXfkPsopLvC5b1Zu3uKXq+vXXsuxL18R/QBemZXFMegwa4oDkGDXVEcgga7ojgEDXZFcQixmo1/8CglTl/8zEWb9qHcUXToWR7Ps3VLF9HbfXMJ0vYe/lX0dhzCY6FGn5P38QTv4DLcqiXjkTb16jxx/YosnEH9uAZnxwHg6nfdSatYMYXorbOYS4Hdzn4keh+e5iz/tDflqw+JizUmbf9f+UgLPHBNXJ/vyV+k3YxXWvTu6MJly5O/+UxwAilbczY8c0a+cgAA/av+TNr29WVIq713m7j+2JSUpK3aJ2f+T/lzw5D1M7i7LQBMTsufpSd5Dorerzfy61u+kew9ccz18/yAJ6H9Bz2zK4pD0GBXFIegwa4oDkGDXVEcQqwm6NzdQpH2jWMu2tUH9USvxz5OyhyZJR/uqsyc3Fq/sarorZ+AE2w9W64UvWfOciIr5RYuzV35a0Jx/cYk1UnzmbNC9B4O4fFAj7b6iN4ZzbiUuERj7poLAFPz8aimJ9+vE733+3An2Ba3mpH2a+j34nqfN3iO+qBBxUTvoaWcmN2XSJ5XvvMEd0y9UF/uzDphCe/n/moal/z6N5X7KPw2nZOfcYKLCE7AfQF3l525c5foHfrBd6QlWCyXdLuf51LiFVN5PQDsSuU6p/5+6N97w/4/emZXFIegwa4oDkGDXVEcgga7ojiE6PSg8zXGXDHGHI6iDTTGXDDGHIj4I2+8VhTlpcGEt33/B4MxJQHcBTAjyqy3gQDuWmvlgVlPu69UKSxqumYLH6dqJXqbvsdND0bOkjtsHlzdmrRC97uJ3tve75EWpxqXeQLAz/5c7hr2JncUvbOa7xMA9njwVYJj5bkRAwB4pz1A2sQPZoreUWUvkpYZZ0SvR4pzpK3uJ3sDzgSQtiohzx07uTGzuL5Xs/ukHczP7w0ArHnUkLQyA7lxBABky81XDw6clxtK/JiNG4bU6vcHaYfWyY0jyg3jWW+fjZJfr333R5L27i250/A9f55N1zaefPVh1Id8ReDUiqOi1w5xPc+GPQjAkyehRvI+6/gnRVFeMWLynb2jMeZQxK/5yZ7bESmK8kJ41mCfBCAbgPwAggCMfpox6qw3hIY+48MpihJTninYrbWXrbWPrbVPAEwBUPgfvP+Z9YYE8qxuRVFePM9ULmuMSRtlimtNAIf/yR9J6sRx0aSI63idc+PkMU3nCuYmrUrCCaK3yTgufTw677ToLfwNJ3aKxT0mOAFvH+7CWmQu7yvONkAuZSzVkhM7edPLo+wH37lKWoXh7qJ34xpOCKavJickWzWvTdqqzv1F7/5P25AWpwwnCQvGE/M/qFiYBwOdXv2p6K31MyfN7lcYK3rH5fiAtCvtz4veIkIX1y9LcdJtyUG530FY+bOk/bqJ98MDQJe53O345lAuGQaA1XlLk7Y1r5zoPDnlc9ImZpNLa/8s6rovf9keeT89EI1gjxj/VBpASmNMIIABAEobY/IjfHprAAD+lCiK8lLxrOOfpr6AY1EU5QWiFXSK4hA02BXFIWiwK4pDiNXmFW6hqZDsaFsXLZ8PlxwCwOFC3BCiU/05ojdJdW6mcObbyqL3eFYuyfRZJf/MW/w+l8a+8SdvA2iaXFyOPom5gUbYzsGit2Tju6Q9XptX9P41k68S7C/eXvT2GtKJtAOb5QYLH/bgGWN1BrUjLX22h+L6JDvY611e7gjs25bLP+8Wl69UZN/CVzXyZ5SbgGSswY0myoV2Ja1l0R7i+lQ9uXz8yZo0orfUVe4u2yU+z5UDgBFHh5LWdJJci5ajFXcKbp5HrlEZndm1/Ppu60eiD9Azu6I4Bg12RXEIGuyK4hA02BXFIcRqgi5F/ARolO0tF+23JPIe9XrXn5DWcIFcLvtLaz/StgTKCasB700kLdXqAqI3TXoeR9R3OmfjPBfLo5eSz+b98IN8aorehUk4YfRDZe4MCwCJ8v5O2q3APaL3WvKvScs3g8taAcAnzTLSNtfmJGX1nvI+7I83JyLNq2g60ZvvwXjSPM7IpaaLVmQjbVPLIaLXuzdXbrermJa0rzfLFd5/XOHS50V15E60Pb59TFr3im+L3oEruQT2zUJLRe/sYVzHdtozUPSu/NC1X8HH94JFH6BndkVxDBrsiuIQNNgVxSFosCuKQ9BgVxSHEKvZ+DsG2PK3fgxBYfIhXA2qRZr/cHlGmVsol6Cu2ZBF9M7v0oe0tA3CRG/iJJdJ29aXO6gerX1bXJ93DDek+PKRtGMYOOvHjXoLJ5a73gY34SsYez6/JHorzeMZcOm7yFn+zoaPN+wy9xrte4kzywCQbDnPWQsZf0f0Vu7DV0CujMsqeoOSe5GWuLfcFblOuaSkVZzGnWQfdJZfA/fljUm7M1L+LHV4j690TFjZSPR+e+MX0urmTSJ6K2biZhtDkspXD7zfdi2XDU0gN2IB9MyuKI5Bg11RHIIGu6I4hOiMf8pojNlkjDlqjDlijPk8Qk9ujFlvjPGP+Ft7xyvKS0x0xj+lBZDWWrvPGOMBYC+AjwE0A3DdWjvCGNMbQDJrba9/uq9E+VLbnL+6JjDyDpC7UN/KxYmSU7vmit5jH3CiY2dPLmUEADOYEx23s8vjfXr98C1pq671Iy1ol/xz7vdCPOJoxoFtghPYuIDHBn1SfpLofVCYWwAeSS3vUb/Sbi9p76SWu7i+e467uPYrxt1ai07zF9cXbzCOtI9Gyl1Rj2WvQ9qMqpxcA4D5F/i1ybqf96gDwO2bnFBssnYVae+7yfvZ9/3Jx2DTNhG9WzJeIO1Sl3ii93NP1gMDeIwXAFyvyonhCV3l8u+EA0Ncbrdq0RfH/E4/8/inIGvtvoh/3wHgByA9gBoApkfYpiP8B4CiKC8p/9N3dmOMN4ACAHYCSB2ld/wlAKmf65EpivJciXawG2OSAPgFQBdrrcuFZRv+XUD8PhB1/FNYMF+jVhQldohWsBtj3BEe6LOttZFVGpcjvs9Hfq+/Iq2NOv4pbgruK6coSuwQnQSdQfh38uvW2i5R9G8ABEdJ0CW31n7xj/eVI66N+6NrMq3OoPdF73fJeOTOt+t53jkAxNnFiaz6GeX93TXf5YqyFL6ciAOAm3vfJG3QMfYuqLlFXH/+4XrSVvaWm2b+EcCNIU/95CZ6P5zJ45u2B1wTvR5hpUjr+6V8v5va8POoEtiZtNWr5Jl9cU+XIG1sEk7aAUDFZVxBF1BGnkG++iMe1ZTgXXnkVpLNvM+96N7E7CvG7y0ArP6a34d8+7xF7/Xy3F+hwFn5uJZ24MRsiUucEAWA1kHctDLpU34pPlLbtRFl1Rb9cegpCbrolMt+AKAJgL+MMQcitL4ARgBYYIxpCeAsgHrRuC9FUf4lojP+6Q8A8iQ/4MPneziKorwotIJOURyCBruiOAQNdkVxCLG6nz3njdyYsdA1Q71znFwuW6C2B2ljO/A4JQCoPoKz8Z5fTRa9aTLyBYMvinFXVQDot4L3qbdvxZ1Vfdxyi+vvbuRyyuIXWsnHVSgVae4N+YoEADxsup20QRnl7qMlfPl44z3lbX9UhVO+u74rT1rKhAPF9U3WcCnxunbynu30HYuS1uDgEtHrtZIz2d8HcedeAJhdmLv3Vo/H++zrXJfHJPlMH8Trx8nlspXi/UbajEC59PrYdxVJe7+7eLUaPRZxiXFCv9Kid+Qu17LyS4E8RiwSPbMrikPQYFcUh6DBrigOQYNdURxCrCboErv54T0P14Rc/Q3viN5pQZxgC9kozzZPcLAsaW6l5f3ObXLy3uYEE7gJJQCEdeQ9yL+PqkRa2VY7xPX+DVKS5r2Pk1gAsHcEH+87C+QE3WZPd9J2z98vemfsyEPasFC5cefV9/i1SfSEX/NOxX3E9V5nOKk6s7k8V3z8gm6sbZMTpUOE6s+d/rI3pAd/nua05b4CcwLmieu9j3Hjzjju1UXvztpcGjvlK26UCgAJh3KJ8jdp+PMBAN17biZtw1T5Nf8mfWuX29uPc/I2Ej2zK4pD0GBXFIegwa4oDkGDXVEcgga7ojiE/9q84nmSNGEuWzzbDBdt+y65IUW+4Tyep/YVLj8FgHdr1SBtvm9L0XttNXdmPblNLnH86C0e2XPEi7vWLmvM5ZgAsGbZQdLyX5J3Ba8Y8JC0oqnLid53/0hD2oDra0Xvpzm42cb69t+L3jI7eHTQ2t73SGt9Uy5bfrstN46YOGKA6K1+hEtCfT/ikmEAON/3d9K848odbvvmf4+09EPOkTbwiNzA40TPnKQ1WVZa9H7oxu9vzcNcngwAWTz4OQy/wc0+AGDOkbyk/fqDaMWNBBNdbo//pj0Cz514tu6yiqK8HmiwK4pD0GBXFIcQk/FPA40xF4wxByL+VHnxh6soyrMSnXLZMADdo45/MsZEbkofa63lweJPIXlWN9Sd6zoqKXhtDtG7w58TOJ9c4L3KAJCmMpe7TtmQQvQmLs3lrmV8/ETvHwG8n73XXJ5NnnlHX3G9x3kuhyzUUy7jPeXHyamOfWaJ3hl9p5EW94CcoEvjzc/h4YkFote9WwhpcWvzc5sYt424foMPP7fcLbjDLgDs68qJYf9+mUVv62FcltreS55z/2Qjjwgbdp0TXucWbBTX+9SsRlqVwGyi9x0PThjvvcCl2wCQOg8PTLrUiBN8ABCWm9/3lgvlgUsz2o13uR03WN4jD0Sv4WQQgKCIf98xxkSOf1IU5RUiJuOfAKCjMeaQMcZXp7gqystNTMY/TQKQDUB+hJ/5Rz9l3X/GP925wRM2FUWJHZ55/JO19rK19rG19gmAKQDEZnJRxz95JEv+vI5bUZT/kehk4w2AqQD8rLVjouhpo9hqAuDB54qivDTEZPxTA2NMfoRPbw0AIKdoo5Ak7DqKB7uWE2Za95Ho3VlvOGlr4nKGHgC+XMIliov39xa9q4ZzNt6jbhHRe+fqWNKOdJ9N2omy4jcY7CjCnT5P5eOMNwC8HXKLtJ7n94neObn48a4Mlxtw1L5ajDS/G5VF7y+9uAHG/QxcijytNZf2AsCoL1lLlqWj6N1ch5tqVFzPM90AoMxffMEnbKJ89aHkHf48uZ/gjrFbH2QR11epx7PpavbML3rTb+bmIt4Z2orenmkzkXbLS54Ll6AEe3+aK8+me1DQNU6enKsr+oCYjX+S3xlFUV5KtIJOURyCBruiOAQNdkVxCLG6nz1lhrdt1Q6+Llqa4C2it9f9vaR1nS2Xj/4+dw9piUYL2SIASR9z8d+FLHK5q1eLm6SlT32UtPuVz4rr7zTnjqDfpPASvVl68YijpZk5iQUAO87y3uhO14eJ3gOleC936V3pRG/JsUNJW+7xAWnzK3JJKgC8+xWPOJrWurjoDcnJnXs3JZD3+tfOwEmnfrnkROfprNwJ9mAdHs+VsHBicf3DbpzArfZZU9E7uHsp0jxvnBe9t0s0Jm3kOf4sAUDN3J1Ja374pOhtctU1Jqr4H8XBkHu6n11RnIwGu6I4BA12RXEIGuyK4hA02BXFIcTqrLeHNhBBYa5lnZuPcpkoAHzTgDPkq1JlkO/4IGfYD8SRS0JHNuXMbrVOG0Rv4lDOJLv9yA0LDs6TO6gOHTqNtG3dOWMNABu8+efu0Zk9RK9HTr7ScDZ+fdFb6t0DpNX0vSx6cz/kuW4enXin4u6dcomzR3W+suK/+Q3Ru/b416Tt+50z1gDwxemCpFU+JpfL5j1Yh7ScA0eSNm2WfFWk/ZVvSEs7Qm4yUWwkv44h1ZaK3jMZuaNvzbvy65jyS+52HKeCXDrd/fRMl9vnH/Dz/899PPV/FEV5rdBgVxSHoMGuKA5Bg11RHEKsJujSJfHEgBKu3TtLVX1X9OaqzUPt03TIKnp9j7Qn7VbI+6J31fZ17GSPIpIAAA5eSURBVF0klwx71vqTtJt7w0gr0S2BuN7tl4Wk7fHi0k0AuDucE4dlt7YTvad2zSFtXA8uawWAM41ykZbRS24X+MvEP0h750cebTV1i/xYFxu9Tdqs5lxSCgClHzUjrUajeqI3/swA0jzvfSV6427pRNqZafx6LUgsdz7P32I8aSPqy8/3XlIuje04m5OcALB7Bu9dT1yfy7EBwPfmKtLW3ZDLob/0c3e57V7l6SGtZ3ZFcQga7IriEDTYFcUhRKfhZAJjzC5jzMGI8U8+EXoWY8xOY8xJY8x8YwzvWVQU5aUhOgm6BwDKWmvvRrSU/sMYsxpAN4SPf5pnjPkBQEuE95J/KqcTpMQnOV3npscpLidaap/nxJ1vhSait0tlHp2UvtBT9hVP573gIaaL6G06m6uRbpflGeY9x8hNBmeD90aHbuRxTADwKNUu0obmvC96W1zgn6vlQ+WftflX82u2e3Rq0bun/QjSVg/mZF6id4qK6+/9NpO0DlPlBoh3Qjk5NeVtd8EJTK/P7Q6Pfizv9b8+MSlpC0MCSUs+eKq4fkVAAdKCD8lVnk2vcOVl8iLya9N0WRBpiztuF71t5nDCuUVZeU99ojybXG4/vvP02Qz/9cxuw4lsk+oe8ccCKAtgUYQ+HYA8jEpRlJeC6A6JcItoI30FwHoApwDctNZGXocKhM5/U5SXmmgFe8Tkl/wAMiB88kvO6D5A1PFPT4KvPeNhKooSU/6nbLy19iaATQCKAvA0xkR+588AgLeDwXX8U5wUPMJYUZTYITrZ+FTGGM+IfycEUB6AH8KDPjKD1RTAshd1kIqixJzoZOPTAphujHFD+A+HBdbalcaYowDmGWOGANiP8Hlw/0jmSyH4buQhF21VHi5PBIBCNbqSNsu7tOg9nGwraXE28/5hADjQnzPcAYfnid4RO8rx+jqNSFvUk8tMAeB0X84i19naT/T2/5P3d09Iwp1hAaD21F6kBa3iPdsAsLc3d4L9MSE/LwC4tngjaSdHcDZ/8BT56kOK6dwBtc5yvlICAH4jeJyRx8lPRe+QkVxy27stvw8A0HDRKdI63/uMtFGTj4jr21fk1NMbJWVvr9Q7SOu2Rd5PXuAJZ9hTHx4iesfH4yszfXI/EL1ebq4l1fuWThZ9QPTGPx1C+Ez2v+un8ZTJrYqivHxoBZ2iOAQNdkVxCBrsiuIQYnX8U9ocyWzLyWVdtGFreK8xALw96wlp45fL+7szr+ek16BknBABgI3v837yt57ITSCTbOKfhX88Tk7a7C/l+e47ijcjrc41ef9+wboNSau46p58vylXkDb2zHTRuys372dPfZL3qANAqTy8bzt1ximkPfTjxBQAdI7Hl1Y7vMcJMwBomZFnvJ9ZISc6g6/zc8vxwQLRm/59njF/u6g/ab+HbBbXv9ecS3ab3DkhehueH0Na3fjycWWpFUpajvPc3BIAQrbu5vXgUmYAuNqwo8vtX6e2QPDFYzr+SVGcjAa7ojgEDXZFcQga7IriEDTYFcUhxGp3WeOfCnHLu2bU19zxEL1uS7lJRKmS3qK3XwCPAkroKf8cy1d/LGktKsglnc1PcwnryUolSMv/uTweqP5ZbnqQvDuXiQLAnWatSavcKaHofecC32+83RNE762r3CAh1zZ57NCGj06TtmgNlzPvmrxcXL+vDHcE7rVKbiLibbgc+vRncia79B3O/k/LLW+83OXF5a53dnUkre2U0eL642O5/8q4LGJyG2OKfE7alJ4/id7kkyaSdnGhPIKq1Pd8JWn8dG6qAQDTF7leQSl8Q7QB0DO7ojgGDXZFcQga7IriEDTYFcUhxGqCLn5mD3gPLO2i/dT8qOhN9CgLaX2+l3fUFhvfhrT+cm4L6bd5ktb5gLfo7fM1J+j2ZviRtLTH5GRPs4w8xqfyu9tEb/PhnCRMV0suS21WkpM1DUrL/T77n+eRTMMXyce7Kl430kJycAfVD5OlEdcX/IHLmRNtk3sFVNz+Hmn3Py0pem812kxakThiYyQc2silxNN+45FQNXryaC4AaDWCk3lh23qK3hkV+bjynOQEMABMDeBuumfSVxa9Zz/l8u+Mp+TEX5curj0Azs/nz1wkemZXFIegwa4oDkGDXVEcQkzGP00zxpwxxhyI+JP/xR+uoijPSkzGPwFAT2vton9YqyjKS0J0Gk5aANL4p/+ZRJ73UaDmXy5aysVyfd+sCdzIoFWRmqK3yUruoLpwz0DR2/wqZ62rdSsrOIETHfmKwIzU/FjHMnH3UwBoNuwxaQmy+oneTLk5C7ytrzzbLrdnRtJWxs0qer+e35c0twtyJvvts5wNb+nhS9rGQ83E9bficdfaJm25LBYAegXxvLnHe5aK3u7v8uM173aXjQD2rOBmHWdCeZZf0eV9xPWBi1uQ1mJQIdGb3msaaWE+1UXvvmQ86y1nS26EAgAT+yYi7c3BM0TvuqWu5ebXb4SIPuAZxz9Za3dG/NdQY8whY8xYY4zc/kRRlJeCZxr/ZIzJDaAPwsdAFQKQHAA3M4fr+Kcbwf9Qpa8oygvlWcc/VbLWBkVMeH0A4Gc8pYd81PFPyVLw+F9FUWKHZx3/dMwYkzZCMwgf13z4RR6ooigxIybjnzYaY1IBMAAOAJBnAkV9MHsSqcOqumi3R/KIJAB43JDLDttNlAfNP/iMR+707c+JJQCo51GGtHR/zhK9KZJyGiLTRC7vvb1dLgn13sz3GzLxB9HbZgDvZ//zxu+iN12Bv0i7ujy76M1cgktgt31QUPTOHHOJtC8PNyYtMG0ecf38N3hf/1nIY7hSL+NEWthwLs0FgPfKrCOt7e6hotd9QV7SenarSpp/yCHSAGDsFh4rFZhFfm07f3qANM9AH9Gbexx3D77VW066pbzKpdN13+oteh/fcT1fr33y9ClsMRn/JKewFUV5KdEKOkVxCBrsiuIQNNgVxSFosCuKQ4jV5hW3bibHqiUNXLT9Z3g+GAA8GsgNKSpckbPxQQf4QsC0ZXLH15659pA26WP5+v/UorX5uDzOklY1f2lxfbA/Z80X+MrllCFfcBb1Yje5scfjApwhH3JMfitbeg8grcp3ciOEbBe/IC3/uNmk3T/Cc/gAYFRm3ibR6S3RipQNuJNsMv9Kore2P5cof/6ZPF+v6c3JpK3uzR1uPQO45BgARp8fQtr1gdNEb/zDPKttf8hl0Xup05ekXXRfI3rLzKlC2uaecinxF31cj+HAd/NFH6BndkVxDBrsiuIQNNgVxSFosCuKQ4jVBF2ioAx4b5hrx9ZB/lwKCQCzknGiZOwoOaHR6khF0hI0KyZ6E+fgfcW7f5f3Kzd+Zzpp94dwmefD6QHi+pOJuaNo9XpcYgkAF7ryCKtJH/DeagBo1Ym962qeEb1mrjtpm/MEi94lGbmbrY93adJWfSv3FaiSZRlpXSY3EJxA1fSPSGvcbJzo7dFlBGmnCvGYJwAYNYS95jSX9zZqJu8lD14aRlqztc1Fb9G2/P6WnXVC9PbIf4W0IUXrit7jqVqSNjfBYtEbf8oSl9txrsnJU0DP7IriGDTYFcUhaLArikPQYFcUh6DBrigOwYQ3j40d0ufKYjvMcC3f9Fsjl4ROz8wb+IM2bRW9TQ9xN87g7fJm/28rcDYeWauJ3oo/cEOJnNMzkDbhe87QA8DoPtxz75M85UXvhgu1SLtf7U3R65uFj6vYDtYAYKIXN1M4MJU7uwJA8yv7SSuRhrO78yq+I65PfpjLi/P1Gih6L+XjEtbAeDlF77kEfKXh/c+4nBoANq4IIG1FpqKktR58TVyf/9P6pH3sw1dlAKBHDe6cu7OgnA3v+skbpM2olFj05jQ8n29TWfkKyts7O7jcbnu4Oo7f+8tIXj2zK4pD0GBXFIegwa4oDkGDXVEcQqwm6IwxVwFEbghPCUDOkrza6PN69Xidnltmay1ntxHLwe7ywMbssdbKPY1fYfR5vXq8zs8tKvprvKI4BA12RXEI/2awc7Ow1wN9Xq8er/Nz+w//2nd2RVFiF/01XlEcQqwHuzGmkjHmuDHmpDFGLmB/RTDG+BpjrhhjDkfRkhtj1htj/CP+fuXmVBtjMhpjNhljjhpjjhhjPo/QX+nnZoxJYIzZZYw5GPG8fCL0LMaYnRGfyfnGmHj/9rG+CGI12CMmwU4AUBlALgANjDHyLpJXg2kA/t7svDeA36y12QH8FnH7VSMMQHdrbS4ARQB0iHifXvXn9gBAWWttPgD5AVQyxhQBMBLAWGvtmwBuAOC+UK8BsX1mLwzgpLX2tLX2IYB5AGrE8jE8N6y1vwP4++SKGgAit0lNR/js+lcKa22QtXZfxL/vAPADkB6v+HOz4dyNuOke8ccCKAsgcsLFK/e8oktsB3t6AFH3NgZGaK8Tqa21kftoLwGQ95O+IhhjvBE+snsnXoPnZoxxM8YcAHAFwHoApwDctNZGdpp8HT+TADRB90Kx4Zc6XtnLHcaYJAB+AdDFWns76v+9qs/NWvvYWpsfQAaE/6Ypb6J/DYntYL8AIOqQrQwR2uvEZWNMWgCI+Jt7CL8CGGPcER7os621kX2MX4vnBgDW2psANgEoCsDTGBPZVv11/EwCiP1g3w0ge0T2Mx6ATwAsj+VjeNEsB9A04t9NAXAz9ZccY4wBMBWAn7V2TJT/eqWfmzEmlTHGM+LfCQGUR3g+YhOAOhG2V+55RZdYL6oxxlQBMA6AGwBfa+3QWD2A54gxZi6A0gjfNXUZwAAASwEsAJAJ4Tv86llr5fGzLynGmOIAtgL4C0Bkn6W+CP/e/so+N2NMXoQn4NwQfqJbYK0dZIzJivBkcXIA+wE0ttY++PeO9MWgFXSK4hA0QacoDkGDXVEcgga7ojgEDXZFcQga7IriEDTYFcUhaLArikPQYFcUh/B/A9WWCIUtsRkAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LdklGhKR42S7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 351
        },
        "outputId": "d4b7b32f-8f28-4784-9fac-9b11f9c5ef7c"
      },
      "source": [
        "som = SOM(data_set)\n",
        "som.train(4, 0.5)\n",
        "plt.imshow(som.weight.astype('uint8'))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0\n",
            "1\n",
            "2\n",
            "3\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f1946c10410>"
            ]
          },
          "metadata": {},
          "execution_count": 4
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "im55-dMmeGAn"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}