doc/tutorial/pipeline/pipeline_tutorial_homo_nn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Pipeline Tutorial"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### install"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`Pipeline` is distributed along with [fate_client](https://pypi.org/project/fate-client/).\n",
    "\n",
    "```bash\n",
    "pip install fate_client\n",
    "```\n",
    "\n",
    "To use Pipeline, we need to first specify which `FATE Flow Service` to connect to. Once `fate_client` installed, one can find an cmd enterpoint name `pipeline`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Usage: pipeline [OPTIONS] COMMAND [ARGS]...\r\n",
      "\r\n",
      "Options:\r\n",
      "  --help  Show this message and exit.\r\n",
      "\r\n",
      "Commands:\r\n",
      "  init  \b - DESCRIPTION: Pipeline Config Command.\r\n"
     ]
    }
   ],
   "source": [
    "!pipeline --help"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Assume we have a `FATE Flow Service` in 127.0.0.1:9380(defaults in standalone), then exec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Pipeline configuration succeeded.\r\n"
     ]
    }
   ],
   "source": [
    "!pipeline init --ip 127.0.0.1 --port 9380"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### homo nn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The `pipeline` package provides components to compose a `FATE pipeline`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pipeline.backend.pipeline import PipeLine\n",
    "from pipeline.component import DataTransform\n",
    "from pipeline.component import Reader\n",
    "from pipeline.component import HomoNN\n",
    "from pipeline.interface import Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Make a `pipeline` instance:\n",
    "\n",
    "    - initiator: \n",
    "        * role: guest\n",
    "        * party: 9999\n",
    "    - roles:\n",
    "        * guest: 9999\n",
    "        * host: [10000, 9999]\n",
    "        * arbiter: 9999\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeline = PipeLine() \\\n",
    "        .set_initiator(role='guest', party_id=9999) \\\n",
    "        .set_roles(guest=9999, host=[10000], arbiter=10000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a `Reader` to load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "reader_0 = Reader(name=\"reader_0\")\n",
    "# set guest parameter\n",
    "reader_0.get_party_instance(role='guest', party_id=9999).component_param(\n",
    "    table={\"name\": \"breast_homo_guest\", \"namespace\": \"experiment\"})\n",
    "# set host parameter\n",
    "reader_0.get_party_instance(role='host', party_id=10000).component_param(\n",
    "    table={\"name\": \"breast_homo_host\", \"namespace\": \"experiment\"})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Add a `DataTransform` component to parse raw data into Data Instance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_transform_0 = DataTransform(name=\"data_transform_0\", with_label=True)\n",
    "# set guest parameter\n",
    "data_transform_0.get_party_instance(role='guest', party_id=9999).component_param(\n",
    "    with_label=True)\n",
    "data_transform_0.get_party_instance(role='host', party_id=[10000]).component_param(\n",
    "    with_label=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we define the `HomoNN` component."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "homo_nn_0 = HomoNN(\n",
    "    name=\"homo_nn_0\", \n",
    "    max_iter=10, \n",
    "    batch_size=-1, \n",
    "    early_stop={\"early_stop\": \"diff\", \"eps\": 0.0001})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Add single `Dense` layer:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<pipeline.component.homo_nn.HomoNN at 0x1025a2a58>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.keras.layers import Dense\n",
    "homo_nn_0.add(\n",
    "    Dense(units=1, input_shape=(10,), activation=\"sigmoid\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compile:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<pipeline.component.homo_nn.HomoNN at 0x1025a2a58>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.keras import optimizers\n",
    "homo_nn_0.compile(\n",
    "    optimizer=optimizers.Adam(learning_rate=0.05), \n",
    "    metrics=[\"accuracy\", \"AUC\"],\n",
    "    loss=\"binary_crossentropy\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Add components to pipeline:\n",
    "\n",
    "    - data_transform_0 comsume reader_0's output data\n",
    "    - homo_nn_0 comsume data_transform_0's output data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipeline.add_component(reader_0)\n",
    "pipeline.add_component(data_transform_0, data=Data(data=reader_0.output.data))\n",
    "pipeline.add_component(homo_nn_0, data=Data(train_data=data_transform_0.output.data))\n",
    "pipeline.compile();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, submit(fit) our pipeline:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2020-11-02 17:39:31.756 | INFO     | pipeline.utils.invoker.job_submitter:monitor_job_status:121 - Job id is 2020110217393142628946\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Job is still waiting, time elapse: 0:00:00\n",
      "Running component reader_0, time elapse: 0:00:02\n",
      "Running component data_transform_0, time elapse: 0:00:04\n",
      "Running component homo_nn_0, time elapse: 0:00:17\r"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2020-11-02 17:39:50.461 | INFO     | pipeline.utils.invoker.job_submitter:monitor_job_status:129 - Job is success!!! Job id is 2020110217393142628946\n",
      "2020-11-02 17:39:50.462 | INFO     | pipeline.utils.invoker.job_submitter:monitor_job_status:130 - Total time: 0:00:18\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\r"
     ]
    }
   ],
   "source": [
    "pipeline.fit()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Success! Now we can get model summary from homo_nn_0:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'is_converged': False,\n",
       " 'loss_history': [0.45709115266799927,\n",
       "  0.3215165436267853,\n",
       "  0.24622784554958344,\n",
       "  0.19835461676120758,\n",
       "  0.16508619487285614,\n",
       "  0.14108890295028687,\n",
       "  0.12376809120178223,\n",
       "  0.11123336106538773,\n",
       "  0.10193527489900589,\n",
       "  0.09480327367782593]}"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "summary = pipeline.get_component(\"homo_nn_0\").get_summary()\n",
    "summary"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we can use the summary data to draw the loss curve:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x14c24f7f0>]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%pylab inline\n",
    "pylab.plot(summary['loss_history'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For more examples about using pipeline to submit `HomoNN` jobs, please refer to [HomoNN Examples](https://github.com/FederatedAI/FATE/tree/master/examples/pipeline/homo_nn)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}