Maximum_likelihood_estimation_and_TensorFlow_Probability.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Maximum likelihood estimation and TensorFlow Probability.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyNhHCr7nVNPeLvXuECfTsZu",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/adventuresinML/adventures-in-ml-code/blob/master/Maximum_likelihood_estimation_and_TensorFlow_Probability.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZyMipbqcsMex"
      },
      "source": [
        "import tensorflow as tf\n",
        "import tensorflow_probability as tfp\n",
        "import numpy as np\n",
        "import matplotlib.pylab as plt\n",
        "tfd = tfp.distributions"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2jFHlKEWtHZg"
      },
      "source": [
        "x_range = np.arange(0, 10, 0.1)\n",
        "grad = 2.0\n",
        "intercept = 3.0\n",
        "lin_reg = x_range * grad + np.random.normal(0, 3.0, len(x_range)) + intercept"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2vwj8twvwmHP",
        "outputId": "d13a096f-4c29-4e6d-d23d-b1c3845b52b3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "plt.scatter(x_range, lin_reg)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "74xnWt8wx7hs"
      },
      "source": [
        "def neg_log_likelihood(y_actual, y_predict):\n",
        "  return -y_predict.log_prob(y_actual)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jWkl84sQw2Zu"
      },
      "source": [
        "model = tf.keras.Sequential([\n",
        "  tf.keras.layers.Dense(1),\n",
        "  tfp.layers.DistributionLambda(lambda x: tfd.Normal(loc=x, scale=1)),\n",
        "])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VFpVcBLXxzvt",
        "outputId": "1e263606-9201-423c-f947-6658e039d530",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "model.compile(optimizer=tf.optimizers.Adam(learning_rate=0.05), loss=neg_log_likelihood)\n",
        "model.fit(x_range, lin_reg, epochs=500, verbose=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7fb04ef76240>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gjiz-h2-yUms"
      },
      "source": [
        "yhat = model(x_range)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xW_fqfjUy3qu"
      },
      "source": [
        "mean = yhat.mean()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ces2-0Ery6LW",
        "outputId": "1e5a0638-98fc-436b-acc7-1b80c1deb3b9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "plt.close(\"all\")\n",
        "plt.scatter(x_range, lin_reg)\n",
        "plt.plot(x_range, mean, label='predicted')\n",
        "plt.plot(x_range, x_range * grad + intercept, label='ground truth')\n",
        "plt.legend(loc=\"upper left\")\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LWCpOYzh30pO"
      },
      "source": [
        "### Estimate std deviation also"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GP8-N3u3741l"
      },
      "source": [
        "def noise(x, grad=0.5, const=2.0):\n",
        "  return np.random.normal(0, grad * x + const)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d5DM9s2W5A3m"
      },
      "source": [
        "x_range = np.arange(0, 10, 0.1)\n",
        "noise = np.array(list(map(noise, x_range)))\n",
        "grad = 2.0\n",
        "intercept = 3.0\n",
        "lin_reg = x_range * grad + intercept + noise"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H2iNjYct8mo1",
        "outputId": "e71e1469-7510-4838-fe84-e529d9471266",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 266
        }
      },
      "source": [
        "plt.scatter(x_range, lin_reg)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2x9KcQ7c3--E"
      },
      "source": [
        "model = tf.keras.Sequential([\n",
        "  tf.keras.layers.Dense(2),\n",
        "  tfp.layers.DistributionLambda(lambda x: tfd.Normal(loc=x[:, 0], scale=1e-3 + tf.math.softplus(0.3 * x[:, 1]))),\n",
        "])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h1BPjVTV4RQx",
        "outputId": "6a19d191-fadc-45dd-f805-984b4706511c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "model.compile(optimizer=tf.optimizers.Adam(), loss=neg_log_likelihood)\n",
        "model.fit(x_range, lin_reg, epochs=1000, verbose=False)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f87be38a588>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 254
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n1MsZ8er4G6_"
      },
      "source": [
        "yhat = model(x_range)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DLaIeFmI4LQH"
      },
      "source": [
        "mean = yhat.mean()\n",
        "upper = mean + 2 * yhat.stddev()\n",
        "lower = mean - 2 * yhat.stddev()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r2nUmC9p4bno",
        "outputId": "e1fa889a-ebe4-4fe6-979c-0ece7009766b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "plt.close(\"all\")\n",
        "plt.scatter(x_range, lin_reg)\n",
        "plt.plot(x_range, mean, label='predicted')\n",
        "plt.fill_between(x_range, lower, upper, alpha=0.1)\n",
        "plt.plot(x_range, x_range * grad + intercept, label='ground truth')\n",
        "plt.legend(loc=\"upper left\")\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9CeNtU0F4rCB",
        "outputId": "8c0c93de-05dd-484c-fe03-2c28a7b7b368",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "model.weights"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Variable 'dense_35/kernel:0' shape=(1, 2) dtype=float32, numpy=array([[2.2931483, 2.4196053]], dtype=float32)>,\n",
              " <tf.Variable 'dense_35/bias:0' shape=(2,) dtype=float32, numpy=array([2.3962705, 2.2277684], dtype=float32)>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 258
        }
      ]
    }
  ]
}