Distance_metrics_compact_v2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# TO DO\n",
    "+ Add non-local hash function √\n",
    "\n",
    "\n",
    "+ add cross validation\n",
    "+ add MNIST\n",
    "+ add IP addresses "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy.stats import spearmanr, pearsonr, kendalltau\n",
    "import pandas as pd\n",
    "import os\n",
    "T = np.transpose\n",
    "from hashfunctions import *\n",
    "os.chdir('/Users/tim/py_code/')\n",
    "from bloom_filters import * \n",
    "import sys\n",
    "sys.path.append('/Users/tim/py_code/FLY_LSH')\n",
    "from read_data import *\n",
    "\n",
    "# import hashlib # might need\n",
    "\n",
    "max_val_hash = sys.maxsize\n",
    "import xxhash\n",
    "import compute_distance_metrics\n",
    "\n",
    "# get MNIST\n",
    "mnist_path = '/Users/tim/py_code/FLY_LSH/mnist10k.txt'\n",
    "mnist = read_generic_data(mnist_path,10000,28*28) # 10 000, 784\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/tim/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:38: RuntimeWarning: invalid value encountered in true_divide\n"
     ]
    }
   ],
   "source": [
    "dat_use = 'MNIST' # Hallem, rand100\n",
    "# dat_use = 'rand100' # 10, 24, 100\n",
    "n_ex = 1000\n",
    "l_ex = 500\n",
    "\n",
    "proj = 'SB4'  # {'DG','SB4','SB5'}\n",
    "dist_met = 'cos' # {'euc','cos'}\n",
    "cor_fun = kendalltau\n",
    "cor_str = '$\\tau$'\n",
    "\n",
    "eps = .01 # [.01, 6E-6]\n",
    "\n",
    "# KC projection parameters\n",
    "# m= 2753 \n",
    "# k= 17\n",
    "# m= 1055\n",
    "# k=6\n",
    "# m = 959\n",
    "# k = 6\n",
    "# m = 350\n",
    "# k = 17\n",
    "# m = 135\n",
    "# k = 7\n",
    "\n",
    "os.chdir('/Users/tim/py_code/FLY_LSH/fly_dat/')\n",
    "dat = pd.read_csv(\"hallem1.txt\",delimiter=' ')\n",
    "if dat_use == 'Hallem':\n",
    "    dat = dat.values\n",
    "elif dat_use[:4] == 'rand':\n",
    "    dIn = int(dat_use[4:])\n",
    "    dat = np.random.randn(l_ex,dIn) # random data\n",
    "elif dat_use == 'MNIST':\n",
    "#     assert(0,'Need To add Functionallity')\n",
    "    dIn = 28*28\n",
    "    dat = mnist[:l_ex,:]\n",
    "    \n",
    "n_odors = np.shape(dat)[0]\n",
    "dat_100 = dat/np.mean(dat,axis=0)*100\n",
    "dat_100 = np.nan_to_num(dat_100)\n",
    "U = dat_100\n",
    "assert(np.shape(U)[0]==l_ex)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# ORN space functions\n",
    "def novelty_dist(S,q):\n",
    "    if dist_met == 'cos':\n",
    "        cs = -(np.max([cos_sim(s,q) for s in S])-1)\n",
    "    elif dist_met == 'cor':\n",
    "        cs = -(np.max([cor_sim(s,q) for s in S])-1)\n",
    "    elif dist_met == 'euc':\n",
    "        cs = np.min([euc_sim(s,q) for s in S])\n",
    "    return cs\n",
    "\n",
    "def cos_sim(a,b):\n",
    "    return np.dot(a,b)/(np.sqrt(np.sum(a**2)) * np.sqrt(np.sum(b**2)))\n",
    "def cor_sim(a,b):\n",
    "    a = a-np.mean(a)\n",
    "    b = b-np.mean(b)\n",
    "    return np.dot(a,b)/(np.sqrt(np.sum(a**2)) * np.sqrt(np.sum(b**2)))  \n",
    "def euc_sim(a,b):\n",
    "    return np.sqrt(np.sum((a-b)**2))\n",
    "def tag_cnt(S,q):\n",
    "    return np.max([np.sum(s[q==1]) for s in S])\n",
    "\n",
    "# hash function\n",
    "def get_hash(v,m,k):\n",
    "    # using v, get k hashes in range [0,m]\n",
    "    val = np.sum(v)\n",
    "    this_hash = (np.array([xxhash.xxh32(str(val+i)).intdigest() for i in range(k)])-mn_hash)/(mx_hash-mn_hash)\n",
    "    return this_hash*m\n",
    "\n",
    "def hash_array(A,m,k):\n",
    "    shp = np.shape(A)\n",
    "    if len(shp)==2:\n",
    "        B = list()\n",
    "        for i in range(shp[0]):\n",
    "            B.append(get_hash(A[i],m,k))\n",
    "    elif len(shp)==3: # recursive approach\n",
    "        B = list()\n",
    "        for i in range(shp[0]):\n",
    "            B.append(hash_array(A[i],m,k))\n",
    "    return  np.array(B).astype(int)\n",
    "\n",
    "# utility functions \n",
    "def min_m(n,eps):\n",
    "    return int(np.ceil(n*np.log2(np.e)*np.log2(1/eps)))\n",
    "def opt_k(m,n):\n",
    "    return int(np.floor(np.log(2)*m/n))\n",
    "def e_eps(m,k,n):\n",
    "    # From Broder&Mitenmacher2004\n",
    "    p = (1.0-1.0/m)**(k*n)\n",
    "    return (1-p)**k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# random selection... old guard\n",
    "# np.shape(U) # 111 odors x 24 ORNs\n",
    "# tag_ind = np.random.choice(n_odors,(n_ex,l_ex),replace=True) # has possibility of overlap\n",
    "# query_ind = np.random.choice(n_odors,n_ex,replace=True)\n",
    "# ind_probe = np.array([np.random.choice(n_odors,l_ex+1,replace=False) for i in range(n_ex)]) # 1000, 21\n",
    "# tag_ind = ind_probe[:,:l_ex]\n",
    "# query_ind = ind_probe[:,-1]\n",
    "\n",
    "# S = U[tag_ind] # n_ex, l_ex, n_ORN\n",
    "# q = U[query_ind] # n_ex, n_ORN\n",
    "# N = [novelty_dist(S[i],q[i]) for i in range(n_ex)]\n",
    "\n",
    "\n",
    "# min_m(110,.01)\n",
    "# opt_k(min_m(100,.01),100) # 6E6\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Crossvalidation\n",
    "# n_fold = 10\n",
    "# reps_per_fold = 10 # n-S shuffles per fold # commented out old guard\n",
    "\n",
    "# f_ind = np.round(np.linspace(0,n_odors,n_fold+1)).astype(int)\n",
    "# grps = [np.arange(f_ind[i],f_ind[i+1] ) for i in range(n_fold) ] # perfect \n",
    "\n",
    "# for i in range(n_fold):\n",
    "#     this_grp = grps[i]\n",
    "#     a = np.arange(n_odors)\n",
    "#     a = np.delete(a,this_grp)\n",
    "#     if 0:\n",
    "#         ind_probe = [np.random.choice(a,l_ex,replace=False) for i in range(len(this_grp))]\n",
    "#         tag_ind_i = np.concatenate([ind_probe for i in range(reps_per_fold)]) # 1:10, repeated 10 x\n",
    "#         query_ind_i = np.concatenate(T([this_grp for i in range(reps_per_fold)]))\n",
    "\n",
    "#     # we have convinience of all groups being the same size\n",
    "#     ind_probe = a\n",
    "#     tag_ind_i = np.concatenate([ind_probe for i in range(len(this_grp))])\n",
    "#     query_ind_i = this_grp\n",
    "    \n",
    "#     tag_ind.append(tag_ind_i)\n",
    "#     query_ind.append(query_ind_i)\n",
    "\n",
    "m = min_m(l_ex,eps)\n",
    "k = opt_k(m,l_ex)\n",
    "\n",
    "# Leave one Odor Out Crossvalidation\n",
    "n_ex = l_ex\n",
    "query_ind = list()\n",
    "tag_ind = list()\n",
    "for i in range(n_ex):\n",
    "    a = np.arange(n_odors)\n",
    "    a = np.delete(a,i)\n",
    "    tag_ind.append(a)\n",
    "    \n",
    "# query_ind = np.concatenate(query_ind)\n",
    "# tag_ind = np.concatenate(tag_ind)\n",
    "query_ind = np.arange(n_ex)\n",
    "tag_ind = np.array(tag_ind)\n",
    "\n",
    "S = U[tag_ind] # n_ex, l_ex, n_ORN\n",
    "q = U[query_ind] # n_ex, n_ORN\n",
    "N = [novelty_dist(S[i],q[i]) for i in range(n_ex)]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MNIST_d1:784_PROJ:SB4_cos_m:720_k:1_eps:5.0e-01  OGDIST:0.25 PROJ: 0.36 0.26, LSHBLOOM: 0.36 0.26, HBLOOM 0.02 0.01\n"
     ]
    }
   ],
   "source": [
    "# import compute_distance_metrics\n",
    "# reload(compute_distance_metrics)\n",
    "# eps_try = [.5,.2,.1,.05,.01,.005,.001,.0001,.00001,.000001]\n",
    "eps_try = np.logspace(np.log10(.5),np.log10(1E-6),15)\n",
    "dm = 'cos'\n",
    "dist_mets = list()\n",
    "for ep in eps_try:\n",
    "    dist_mets.append(compute_distance_metrics.get_distance_metrics(S,q,eps=ep,app_str=dat_use,\n",
    "                                                                   dist_met=dm))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEaCAYAAAD+E0veAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a24bd2f28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(-np.log10(eps_try),T(dist_mets)[0],label='Proj Space')\n",
    "plt.plot(-np.log10(eps_try),T(dist_mets)[1],label='LSH Bloom Filter')\n",
    "plt.plot(-np.log10(eps_try),T(dist_mets)[2],label='Hash Bloom Filter')\n",
    "# plt.semilogx(eps_try,T(dist_mets)[0],label='Proj Space')\n",
    "# plt.semilogx(eps_try,T(dist_mets)[1],label='LSH Bloom Filter')\n",
    "# plt.semilogx(eps_try,T(dist_mets)[2],label='Hash Bloom Filter')\n",
    "plt.xlabel('$-Log_{10}$ $\\epsilon$')\n",
    "plt.ylabel('Corr ORN Distance(%s)' %dm)\n",
    "plt.legend()\n",
    "plt.title('%s, (N=%d)' %(dat_use,l_ex))\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dIn = np.shape(S)[2]\n",
    "if proj == 'DG':\n",
    "    M = np.random.randn(dIn,m)\n",
    "elif proj[:2]=='SB':\n",
    "    M = T(create_projection_matrix(dIn,m,proj))\n",
    "else:\n",
    "    assert(0)\n",
    "\n",
    "Sm = np.matmul(S,M) # n_ex, l_ex, m\n",
    "qm = np.matmul(q,M) # n_ex, m\n",
    "Sm_LSH = np.argpartition(Sm,-k)[:,:,-k:] # grab the k largest values\n",
    "qm_LSH = np.argpartition(qm,-k)[:,-k:]\n",
    "\n",
    "Sm_LSH_tag = np.zeros(np.shape(Sm))\n",
    "for i,ii in np.ndenumerate(Sm_LSH):      # produce Boolean tags\n",
    "    Sm_LSH_tag[i[0],i[1],ii] = 1   \n",
    "    \n",
    "qm_LSH_tag = np.zeros(np.shape(qm))\n",
    "for i,ii in enumerate(qm_LSH):\n",
    "     qm_LSH_tag[i,ii] = 1\n",
    "        \n",
    "# bloom filter\n",
    "Sm_LSH_bloom = np.sum(Sm_LSH_tag,axis=1)>0 \n",
    "avg_hash_loc = np.mean(Sm_LSH_bloom,axis=0)\n",
    "\n",
    "# closest hash\n",
    "N_LSH_tag = [-(tag_cnt(Sm_LSH_tag[i],qm_LSH_tag[i])/k-1) for i in range(n_ex)]      # Normalized\n",
    "N_LSH_bloom = -(np.sum([Sm_LSH_bloom[i,qm_LSH[i]] for i in range(n_ex)],axis=1)/k-1)\n",
    "\n",
    "# N_fly = np.zeros(n_ex)\n",
    "# for i in range(n_ex):\n",
    "#     N_fly[i] = fly_bloom(S=Sm_tag[i],q=qm_tag[i],rr=.01)\n",
    "\n",
    "# Non-local hash\n",
    "Sm_hash = hash_array(S,m,k)\n",
    "qm_hash = hash_array(q,m,k)\n",
    "\n",
    "Sm_hash_tag = np.zeros(np.shape(Sm))\n",
    "for i,ii in np.ndenumerate(Sm_hash):      # produce Boolean tags\n",
    "    Sm_hash_tag[i[0],i[1],ii] = 1   \n",
    "    \n",
    "qm_hash_tag = np.zeros(np.shape(qm))\n",
    "for i,ii in enumerate(qm_hash):\n",
    "     qm_hash_tag[i,ii] = 1\n",
    "        \n",
    "# non-local bloom filter\n",
    "Sm_hash_bloom = np.sum(Sm_hash_tag,axis=1)>0 \n",
    "avg_hash_loc = np.mean(Sm_hash_bloom,axis=0)\n",
    "\n",
    "# closest hash\n",
    "N_hash_tag = [-(tag_cnt(Sm_hash_tag[i],qm_hash_tag[i])/k-1) for i in range(n_ex)]      # Normalized\n",
    "N_hash_bloom = -(np.sum([Sm_hash_bloom[i,qm_hash[i]] for i in range(n_ex)],axis=1)/k-1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmQAAAKUCAYAAACqgySlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ns4K5Q2ZmZmZWMHfIzMzMzAr2fwHY6MddMzZrIQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112ddc320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# variables to compare \n",
    "# - N       | ORN correlation\n",
    "# - N_tag   | KC hamming distance\n",
    "# - N_bloom | Bloom filter\n",
    "# - N_fly   | Fly Bloom\n",
    "\n",
    "plt.figure(figsize=(10,10))\n",
    "# plt.suptitle(dat_use + ' [' + proj + ' projection]' '(m)')\n",
    "plt.suptitle('%s [%s projection] (m:%d, k:%d) %s dist' %(dat_use,proj,m,k,dist_met))\n",
    "plt.subplot(2,2,1)\n",
    "plt.hist(N,20,histtype='step',edgecolor='k')\n",
    "plt.xlabel('Novelty (' + dist_met + ')')\n",
    "plt.ylabel('# TAGS')\n",
    "plt.title('ORN Space (n=%d) [d1=%d]' %(l_ex,np.shape(U)[1]))\n",
    "\n",
    "plt.subplot(2,2,2)\n",
    "plt.plot(N,N_LSH_tag,'k.')\n",
    "\n",
    "plt.title( '(m: %i) r = %.2f, $r_T$ = %.2f' %((m,) + (spearmanr(N,N_LSH_tag)[0],) + (kendalltau(N,N_LSH_tag)[0],)))\n",
    "# plt.title( '(m: %i) r = %.2f, p = %.1e' %((m,) + cor_fun(N,N_tag)))\n",
    "plt.ylabel('Novelty KC Space')\n",
    "plt.xlabel('Novelty ORN Space')\n",
    "\n",
    "plt.subplot(2,2,3)\n",
    "plt.plot(N,N_LSH_bloom,'k.')\n",
    "plt.title( '(m: %i) r = %.2f, $r_T$ = %.2f' %((m,) + (spearmanr(N,N_LSH_bloom)[0],) + (kendalltau(N,N_LSH_bloom)[0],)))\n",
    "# plt.title('(m: %i) r = %.2f, p = %.1e' %((m,) + cor_fun(N,N_bloom)))\n",
    "plt.ylabel('Novelty Distance Bloom')\n",
    "plt.xlabel('Novelty ORN Space')\n",
    "\n",
    "plt.subplot(2,2,4)\n",
    "plt.plot(N,N_hash_bloom,'k.')\n",
    "plt.title( '(m: %i) r = %.2f, $r_T$ = %.2f' %((m,) + (spearmanr(N,N_hash_bloom)[0],) + (kendalltau(N,N_hash_bloom)[0],)))\n",
    "# plt.title('(m: %i) r = %.2f, p = %.1e' %((m,) + cor_fun(N,N_fly)))\n",
    "plt.ylabel('Novelty Hash')\n",
    "plt.xlabel('Novelty ORN Space')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.52000000000000002,\n",
       " 0.41999999999999998,\n",
       " 0.42999999999999999,\n",
       " 0.32000000000000001,\n",
       " 0.10000000000000001,\n",
       " 0.070000000000000007)"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def cor_fun(x1,x2):\n",
    "    a1 = pearsonr(x1,x2)[0]\n",
    "    a2 = kendalltau(x1,x2)[0]\n",
    "    return np.round([a1,a2],2)\n",
    "\n",
    "\n",
    "m_0 = '%.2f' %np.mean(N) # dist ORN space\n",
    "m_1 = cor_fun(N,N_LSH_tag)\n",
    "m_2 = cor_fun(N,N_LSH_bloom)\n",
    "m_3 = cor_fun(N,N_hash_bloom)\n",
    "foo = np.concatenate([m_1, m_2, m_3])\n",
    "mPRT = 'PROJ: %.2f %.2f, LSHBLOOM: %.2f %.2f, HBLOOM %.2f %.2f' %tuple(np.concatenate([m_1, m_2, m_3]))\n",
    "tuple(foo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "' 5.05  2.    3.  '"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "foo = np.round((5.050001 ,2, 3),2)\n",
    "str(foo)[1:-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "foo = '%.1e' %.005\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'PROJ: 0.52 0.42, LSHBLOOM: 0.43 0.32, HBLOOM 0.10 0.07'"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mPRT"
   ]
  },
  {
   "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": "3.4.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}