notebooks/The Helium Demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Imports and Convenience Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "import numpy as np\n",
    "import phe as paillier\n",
    "from sonar.contracts import ModelRepository,Model\n",
    "from sonar.ipfs import IPFS\n",
    "from syft.he.paillier.keys import KeyPair\n",
    "from syft.nn.linear import LinearClassifier\n",
    "from sklearn.datasets import load_diabetes\n",
    "\n",
    "def get_balance(account):\n",
    "    return repo.web3.fromWei(repo.web3.eth.getBalance(account),'ether')\n",
    "\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# this is the addresas\n",
    "my_wallet_address = '0xc6E1D4501D3f354E89F048F096db4424168682a9'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setting up the Experiment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Connected to OpenMined ModelRepository:0xd60e1a150b59a89a8e6e6ff2c03ffb6cb4096205\n"
     ]
    }
   ],
   "source": [
    "# for the purpose of the simulation, we're going to split our dataset up amongst\n",
    "# the relevant simulated users\n",
    "\n",
    "diabetes = load_diabetes()\n",
    "y = diabetes.target\n",
    "X = diabetes.data\n",
    "\n",
    "validation = (X[0:5],y[0:5])\n",
    "anonymous_diabetes_users = (X[6:],y[6:])\n",
    "\n",
    "# we're also going to initialize the model trainer smart contract, which in the\n",
    "# real world would already be on the blockchain (managing other contracts) before\n",
    "# the simulation begins\n",
    "\n",
    "andreas_repo = '0xd60e1a150b59a89a8e6e6ff2c03ffb6cb4096205'\n",
    "stefan_repo = '0xf0DE3D665D777d12609D0524c89C6456A6658E5E'\n",
    "\n",
    "\n",
    "\n",
    "# ATTENTION: copy paste the correct address (NOT THE DEFAULT SEEN HERE) from truffle migrate output.\n",
    "repo = ModelRepository(andreas_repo, account=my_wallet_address,ipfs=IPFS(host='127.0.0.1'), web3_host='localhost') # blockchain hosted model repository\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 1: Cure Diabetes Inc Initializes a Model and Provides a Bounty"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "pubkey,prikey = KeyPair().generate(n_length=1024)\n",
    "diabetes_classifier = LinearClassifier(desc=\"DiabetesClassifier\",n_inputs=10,n_labels=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "initial_error = diabetes_classifier.evaluate(validation[0],validation[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Linear Model (10,1): DiabetesClassifier"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "diabetes_classifier.encrypt(pubkey)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "diabetes_model = Model(owner='0xc6E1D4501D3f354E89F048F096db4424168682a9',\n",
    "                       syft_obj = diabetes_classifier,\n",
    "                       bounty = 0.00000000001,\n",
    "                       initial_error = initial_error,\n",
    "                       target_error = 10000\n",
    "                      )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_id = repo.submit_model(diabetes_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "11"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(repo)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 2: An Anonymous Patient Downloads the Model and Improves It"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_id = 10\n",
    "model = repo[model_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "repo[model_id].submit_gradient(my_wallet_address,anonymous_diabetes_users[0][0],anonymous_diabetes_users[0][0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 3: Cure Diabetes Inc. Evaluates the Gradient "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Desc:DiabetesClassifier\n",
       "Owner:0xc6e1d4501d3f354e89f048f096db4424168682a9\n",
       "Bounty:9.999999E-12\n",
       "Initial Error:21802517\n",
       "Best Error:None\n",
       "Target Error:10000\n",
       "Model ID:10\n",
       "Num Grads:0"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "repo[model_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3.709448782750000015\n"
     ]
    }
   ],
   "source": [
    "old_balance = get_balance(my_wallet_address)\n",
    "print(old_balance)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "grad = model[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grad.id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = repo[model_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_error = model.evaluate_gradient(my_wallet_address,repo[model_id][0],prikey,pubkey,validation[0],validation[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "21802481"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-0.001333360000000000\n"
     ]
    }
   ],
   "source": [
    "new_balance = get_balance(my_wallet_address)\n",
    "incentive = new_balance - old_balance\n",
    "print(incentive)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 4: Rinse and Repeat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Desc:DiabetesClassifier\n",
       "Owner:0xc6e1d4501d3f354e89f048f096db4424168682a9\n",
       "Bounty:9.999999E-12\n",
       "Initial Error:21802517\n",
       "Best Error:None\n",
       "Target Error:10000\n",
       "Model ID:10\n",
       "Num Grads:1"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for i,(addr, input, target) in enumerate(anonymous_diabetics):\n",
    "#     try:\n",
    "        \n",
    "#         model = repo[model_id]\n",
    "        \n",
    "#         # patient is doing this\n",
    "#         model.submit_gradient(addr,input,target)\n",
    "        \n",
    "#         # Cure Diabetes Inc does this\n",
    "#         old_balance = get_balance(addr)\n",
    "#         new_error = model.evaluate_gradient(cure_diabetes_inc,model[i+1],prikey,pubkey,validation[0],validation[1],alpha=2)\n",
    "#         print(\"new error = \"+str(new_error))\n",
    "#         incentive = round(get_balance(addr) - old_balance,5)\n",
    "#         print(\"incentive = \"+str(incentive))\n",
    "#     except:\n",
    "#         \"Connection Reset\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}