ls1 and ls3 svd and nmf

Phase 2/task_9.ipynb

@@ -2,9 +2,18 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": 62,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "The autoreload extension is already loaded. To reload it, use:\n",
+      "  %reload_ext autoreload\n"
+     ]
+    }
+   ],
    "source": [
     "%load_ext autoreload\n",
     "%autoreload 2"
@@ -12,18 +21,31 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 63,
    "metadata": {},
    "outputs": [],
    "source": [
     "import json\n",
     "import os\n",
-    "from utils import *"
+    "import numpy as np\n",
+    "from utils import *\n",
+    "import math\n",
+    "import heapq"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 64,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "fd_collection = getCollection(\"team_5_mwdb_phase_2\", \"fd_collection\")\n",
+    "all_images = fd_collection.find()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 65,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -50,6 +72,11 @@
     "if label < 0 and label > 100:\n",
     "    raise ValueError(\"k should be between 0 and 100\")\n",
     "\n",
+    "knum = int(input(\"Enter value of knum: \"))\n",
+    "if knum < 1:\n",
+    "    raise ValueError(\"knum should be a positive integer\")\n",
+    "\n",
+    "label_rep = calculate_label_representatives(fd_collection, label, selected_feature_model)\n",
     "\n",
     "match selected_latent_space:\n",
     "    case \"\":\n",
@@ -72,34 +99,169 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 66,
    "metadata": {},
    "outputs": [],
-   "source": []
+   "source": [
+    "def extract_similarities_ls1(dim_reduction, data, label, label_rep):\n",
+    "\n",
+    "  match dim_reduction:\n",
+    "\n",
+    "    case 'svd':\n",
+    "      U = np.array(data[\"image-semantic\"])\n",
+    "      S = np.array(data[\"semantics-core\"])\n",
+    "      V = np.transpose(np.array(data[\"semantic-feature\"]))\n",
+    "\n",
+    "      comparison_feature_space = np.matmul(U, S)\n",
+    "      comparison_vector = np.matmul(np.matmul(label_rep, V), S)\n",
+    "    \n",
+    "    case \"nmf\":\n",
+    "      H = np.array(data['semantic-feature'])\n",
+    "      comparison_feature_space = np.array(data['image-semantic'])\n",
+    "      comparison_vector = np.matmul(label_rep, np.transpose(H))\n",
+    "\n",
+    "  print(comparison_feature_space.shape)\n",
+    "  n = len(comparison_feature_space)\n",
+    " \n",
+    "  distances = []\n",
+    "  for i in range(n):\n",
+    "    if i != label:\n",
+    "      distances.append({\"image_id\": i, \"label\": all_images[i][\"true_label\"],\"distance\": math.dist(comparison_vector, comparison_feature_space[i])})\n",
+    "\n",
+    "  distances = sorted(distances, key=lambda x: x[\"distance\"], reverse=False)\n",
+    "\n",
+    "  similar_labels = []\n",
+    "  unique_labels = set()\n",
+    "\n",
+    "  for img in distances:\n",
+    "    if img['label'] not in unique_labels:\n",
+    "      similar_labels.append(img)\n",
+    "      unique_labels.add(img[\"label\"])\n",
+    "\n",
+    "      if len(similar_labels) == knum:\n",
+    "        break\n",
+    "\n",
+    "\n",
+    "  for x in similar_labels:\n",
+    "    print(x)"
+   ]
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 67,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def extract_similarities_ls3(dim_reduction, data, label):\n",
+    "\n",
+    "  match dim_reduction:\n",
+    "\n",
+    "    case 'svd':\n",
+    "      U = np.array(data[\"image-semantic\"])\n",
+    "      S = np.array(data[\"semantics-core\"])\n",
+    "      V = np.transpose(np.array(data[\"semantic-feature\"]))\n",
+    "\n",
+    "      comparison_feature_space = np.matmul(U, S)\n",
+    "      comparison_vector = comparison_feature_space[label]\n",
+    "    \n",
+    "    case \"nmf\":\n",
+    "      comparison_feature_space = np.array(data['image-semantic'])\n",
+    "      comparison_vector = comparison_feature_space[label]\n",
+    "\n",
+    "  n = len(comparison_feature_space)\n",
+    "  distances = []\n",
+    "  for i in range(n):\n",
+    "    if i != label:\n",
+    "      distances.append({\"label\": i, \"distance\": math.dist(comparison_vector, comparison_feature_space[i])})\n",
+    "\n",
+    "  distances = sorted(distances, key=lambda x: x[\"distance\"], reverse=False)[:knum]\n",
+    "\n",
+    "  for x in distances:\n",
+    "    print(x)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 68,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "(101, 10)\n",
+      "{'label': 4, 'distance': 0.9931105104385977}\n",
-      "(10, 10)\n",
+      "{'label': 92, 'distance': 1.1209182190288185}\n",
-      "(10, 101)\n"
+      "{'label': 65, 'distance': 1.2107732156271573}\n",
+      "{'label': 21, 'distance': 1.5053484881391492}\n",
+      "{'label': 2, 'distance': 1.698430977110922}\n",
+      "{'label': 100, 'distance': 1.8636096001573115}\n",
+      "{'label': 95, 'distance': 2.003755992104511}\n",
+      "{'label': 11, 'distance': 2.069066281581252}\n",
+      "{'label': 60, 'distance': 2.070894540798742}\n",
+      "{'label': 88, 'distance': 2.0925931256031}\n",
+      "{'label': 43, 'distance': 2.1056747598887218}\n",
+      "{'label': 33, 'distance': 2.165431005806523}\n",
+      "{'label': 90, 'distance': 2.174626607979455}\n",
+      "{'label': 83, 'distance': 2.188609736988739}\n",
+      "{'label': 68, 'distance': 2.209562202827548}\n",
+      "{'label': 59, 'distance': 2.27130902508622}\n",
+      "{'label': 35, 'distance': 2.276916489521396}\n",
+      "{'label': 70, 'distance': 2.283111150497479}\n",
+      "{'label': 53, 'distance': 2.2871296343421075}\n",
+      "{'label': 42, 'distance': 2.2943393449254192}\n",
+      "{'label': 1, 'distance': 2.299515307388396}\n",
+      "{'label': 89, 'distance': 2.300444335700286}\n",
+      "{'label': 64, 'distance': 2.3105619552648906}\n",
+      "{'label': 47, 'distance': 2.3258018764464126}\n",
+      "{'label': 28, 'distance': 2.33793138436563}\n",
+      "{'label': 91, 'distance': 2.348432279582375}\n",
+      "{'label': 66, 'distance': 2.378823252101462}\n",
+      "{'label': 52, 'distance': 2.3845656934663344}\n",
+      "{'label': 17, 'distance': 2.3851103284430946}\n",
+      "{'label': 29, 'distance': 2.392106657184808}\n",
+      "{'label': 46, 'distance': 2.4059349825734024}\n",
+      "{'label': 98, 'distance': 2.425981349727766}\n",
+      "{'label': 12, 'distance': 2.4320238781945878}\n",
+      "{'label': 5, 'distance': 2.433658250868235}\n",
+      "{'label': 72, 'distance': 2.4438014606638965}\n",
+      "{'label': 96, 'distance': 2.446857205149324}\n",
+      "{'label': 18, 'distance': 2.4473786634019508}\n",
+      "{'label': 0, 'distance': 2.4482053195868017}\n",
+      "{'label': 49, 'distance': 2.451590137889849}\n",
+      "{'label': 14, 'distance': 2.4717097207497414}\n",
+      "{'label': 85, 'distance': 2.473715190942228}\n",
+      "{'label': 19, 'distance': 2.4754273396104534}\n",
+      "{'label': 51, 'distance': 2.4810475345400316}\n",
+      "{'label': 75, 'distance': 2.4850838216864224}\n",
+      "{'label': 93, 'distance': 2.4867224184341175}\n",
+      "{'label': 44, 'distance': 2.498509815319209}\n",
+      "{'label': 82, 'distance': 2.501339416798757}\n",
+      "{'label': 54, 'distance': 2.506342353975533}\n",
+      "{'label': 9, 'distance': 2.5065630929096394}\n",
+      "{'label': 41, 'distance': 2.51345667730748}\n"
      ]
     }
    ],
    "source": [
     "match selected_latent_space:\n",
     "\n",
+    "  case \"\":\n",
+    "    \n",
+    "    extract_similarities_ls1(selected_dim_reduction_method, data, label, label_rep)\n",
+    "\n",
     "  case \"label_sim\":\n",
     "\n",
-    "    extract_simila\n"
+    "    extract_similarities_ls3(selected_dim_reduction_method, data, label)\n",
+    "    "
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
   {
    "cell_type": "code",
    "execution_count": null,

Phase 2/utils.py

@@ -841,6 +841,25 @@ def svd(matrix, k):
 
     return left_singular_vectors, np.diag(singular_values), right_singular_vectors.T
 
+def nmf(matrix, k, num_iterations=100):
+    d1, d2 = matrix.shape
+    # Initialize W and H matrices with random non-negative values
+    W = np.random.rand(d1, k)
+    H = np.random.rand(k, d2)
+
+    for iteration in range(num_iterations):
+        # Update H matrix
+        numerator_h = np.dot(W.T, matrix)
+        denominator_h = np.dot(np.dot(W.T, W), H)
+        H *= numerator_h / denominator_h
+
+        # Update W matrix
+        numerator_w = np.dot(matrix, H.T)
+        denominator_w = np.dot(W, np.dot(H, H.T))
+        W *= numerator_w / denominator_w
+
+    return W, H
+
 def extract_latent_semantics_from_feature_model(
     fd_collection,
     k,
@@ -1087,8 +1106,7 @@ def extract_latent_semantics_from_sim_matrix(
             )
             model.fit(feature_vectors_shifted)
 
-            W = model.transform(feature_vectors_shifted)
+            W, H = nmf(feature_vectors_shifted, k = k)
-            H = model.components_
 
             all_latent_semantics = {
                 "image-semantic": W.tolist(),