1059 lines
22 KiB
Plaintext
1059 lines
22 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"\n",
|
|
"<div class=\"alert alert-block alert-info\">\n",
|
|
"<h1> Ekstrakcja informacji </h1>\n",
|
|
"<h2> 8. <i>Regresja logistyczna</i> [ćwiczenia]</h2> \n",
|
|
"<h3> Jakub Pokrywka (2021)</h3>\n",
|
|
"</div>\n",
|
|
"\n",
|
|
""
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"# Regresja logistyczna"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## import bibliotek"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"import numpy as np\n",
|
|
"import gensim\n",
|
|
"import torch\n",
|
|
"import pandas as pd\n",
|
|
"from sklearn.model_selection import train_test_split\n",
|
|
"\n",
|
|
"from sklearn.datasets import fetch_20newsgroups\n",
|
|
"# https://scikit-learn.org/0.19/datasets/twenty_newsgroups.html\n",
|
|
"\n",
|
|
"from sklearn.feature_extraction.text import TfidfVectorizer\n",
|
|
"from sklearn.metrics import accuracy_score"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"CATEGORIES = ['soc.religion.christian', 'alt.atheism']"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"newsgroups_train_dev = fetch_20newsgroups(subset = 'train', categories=CATEGORIES)\n",
|
|
"newsgroups_test = fetch_20newsgroups(subset = 'test', categories=CATEGORIES)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"newsgroups_train_dev_text = newsgroups_train_dev['data']\n",
|
|
"newsgroups_test_text = newsgroups_test['data']"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 5,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"Y_train_dev = newsgroups_train_dev['target']\n",
|
|
"Y_test = newsgroups_test['target']"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 6,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"newsgroups_train_text, newsgroups_dev_text, Y_train, Y_dev = train_test_split(newsgroups_train_dev_text, Y_train_dev, random_state=42)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 7,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"Y_names = newsgroups_train_dev['target_names']"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"['alt.atheism', 'soc.religion.christian']"
|
|
]
|
|
},
|
|
"execution_count": 8,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"Y_names"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## baseline"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## zadanie (5 minut)\n",
|
|
"\n",
|
|
"- stworzyć baseline "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### PYTANIE: co jest nie tak z regresją liniową?"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Regresja logistyczna"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### wektoryzacja"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## zadanie (5 minut)\n",
|
|
"\n",
|
|
"- na podstawie newsgroups_train_text stworzyć tfidf wektoryzer ze słownikiem max 10_000\n",
|
|
"- wygenerować wektory: X_train, X_dev, X_test"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### model - inicjalizacja "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 18,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"class LogisticRegressionModel(torch.nn.Module):\n",
|
|
"\n",
|
|
" def __init__(self):\n",
|
|
" super(LogisticRegressionModel, self).__init__()\n",
|
|
" self.fc = torch.nn.Linear(FEAUTERES,1)\n",
|
|
"\n",
|
|
" def forward(self, x):\n",
|
|
" x = self.fc(x)\n",
|
|
" x = torch.sigmoid(x)\n",
|
|
" return x"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 19,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"lr_model = LogisticRegressionModel()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 20,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"tensor([[0.4978],\n",
|
|
" [0.5009],\n",
|
|
" [0.4998],\n",
|
|
" [0.4990],\n",
|
|
" [0.5018]], grad_fn=<SigmoidBackward>)"
|
|
]
|
|
},
|
|
"execution_count": 20,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"lr_model(torch.Tensor(X_train[0:5].astype(np.float32).todense()))"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 21,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"LogisticRegressionModel(\n",
|
|
" (fc): Linear(in_features=10000, out_features=1, bias=True)\n",
|
|
")"
|
|
]
|
|
},
|
|
"execution_count": 21,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"lr_model"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 22,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"[Parameter containing:\n",
|
|
" tensor([[-0.0059, 0.0035, 0.0021, ..., -0.0042, -0.0057, -0.0049]],\n",
|
|
" requires_grad=True),\n",
|
|
" Parameter containing:\n",
|
|
" tensor([-0.0023], requires_grad=True)]"
|
|
]
|
|
},
|
|
"execution_count": 22,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"list(lr_model.parameters())"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"## model - trenowanie"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 23,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"BATCH_SIZE = 5"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 24,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"criterion = torch.nn.BCELoss()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 25,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"optimizer = torch.optim.SGD(lr_model.parameters(), lr = 0.1)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 26,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"809"
|
|
]
|
|
},
|
|
"execution_count": 26,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"Y_train.shape[0]"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 27,
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"loss_score = 0\n",
|
|
"acc_score = 0\n",
|
|
"items_total = 0\n",
|
|
"lr_model.train()\n",
|
|
"for i in range(0, Y_train.shape[0], BATCH_SIZE):\n",
|
|
" X = X_train[i:i+BATCH_SIZE]\n",
|
|
" X = torch.tensor(X.astype(np.float32).todense())\n",
|
|
" Y = Y_train[i:i+BATCH_SIZE]\n",
|
|
" Y = torch.tensor(Y.astype(np.float32)).reshape(-1,1)\n",
|
|
" Y_predictions = lr_model(X)\n",
|
|
" acc_score += torch.sum((Y_predictions > 0.5) == Y).item()\n",
|
|
" items_total += Y.shape[0] \n",
|
|
" \n",
|
|
" optimizer.zero_grad()\n",
|
|
" loss = criterion(Y_predictions, Y)\n",
|
|
" loss.backward()\n",
|
|
" optimizer.step()\n",
|
|
" \n",
|
|
"\n",
|
|
" loss_score += loss.item() * Y.shape[0] "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 28,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"tensor([[0.5667],\n",
|
|
" [0.5802],\n",
|
|
" [0.5757],\n",
|
|
" [0.5670]], grad_fn=<SigmoidBackward>)"
|
|
]
|
|
},
|
|
"execution_count": 28,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"Y_predictions"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 29,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"tensor([[0.],\n",
|
|
" [1.],\n",
|
|
" [1.],\n",
|
|
" [0.]])"
|
|
]
|
|
},
|
|
"execution_count": 29,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"Y"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 30,
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"452"
|
|
]
|
|
},
|
|
"execution_count": 30,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"acc_score"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 31,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"809"
|
|
]
|
|
},
|
|
"execution_count": 31,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"items_total"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 32,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"accuracy: 0.5587144622991347\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"print(f'accuracy: {acc_score / items_total}')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 33,
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"BCE loss: 0.6745463597170355\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"print(f'BCE loss: {loss_score / items_total}')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### model - ewaluacja"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 34,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"def get_loss_acc(model, X_dataset, Y_dataset):\n",
|
|
" loss_score = 0\n",
|
|
" acc_score = 0\n",
|
|
" items_total = 0\n",
|
|
" model.eval()\n",
|
|
" for i in range(0, Y_dataset.shape[0], BATCH_SIZE):\n",
|
|
" X = X_dataset[i:i+BATCH_SIZE]\n",
|
|
" X = torch.tensor(X.astype(np.float32).todense())\n",
|
|
" Y = Y_dataset[i:i+BATCH_SIZE]\n",
|
|
" Y = torch.tensor(Y.astype(np.float32)).reshape(-1,1)\n",
|
|
" Y_predictions = model(X)\n",
|
|
" acc_score += torch.sum((Y_predictions > 0.5) == Y).item()\n",
|
|
" items_total += Y.shape[0] \n",
|
|
"\n",
|
|
" loss = criterion(Y_predictions, Y)\n",
|
|
"\n",
|
|
" loss_score += loss.item() * Y.shape[0] \n",
|
|
" return (loss_score / items_total), (acc_score / items_total)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 35,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.6443227143826974, 0.622991347342398)"
|
|
]
|
|
},
|
|
"execution_count": 35,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"get_loss_acc(lr_model, X_train, Y_train)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 36,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.6369243131743537, 0.6037037037037037)"
|
|
]
|
|
},
|
|
"execution_count": 36,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"get_loss_acc(lr_model, X_dev, Y_dev)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 37,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.6323775731785694, 0.6499302649930265)"
|
|
]
|
|
},
|
|
"execution_count": 37,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"get_loss_acc(lr_model, X_test, Y_test)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### wagi modelu"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 38,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"[Parameter containing:\n",
|
|
" tensor([[ 0.0314, -0.0375, 0.0131, ..., -0.0057, -0.0008, -0.0089]],\n",
|
|
" requires_grad=True),\n",
|
|
" Parameter containing:\n",
|
|
" tensor([0.0563], requires_grad=True)]"
|
|
]
|
|
},
|
|
"execution_count": 38,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"list(lr_model.parameters())"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 39,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"tensor([ 0.0314, -0.0375, 0.0131, ..., -0.0057, -0.0008, -0.0089],\n",
|
|
" grad_fn=<SelectBackward>)"
|
|
]
|
|
},
|
|
"execution_count": 39,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"list(lr_model.parameters())[0][0]"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 40,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"torch.return_types.topk(\n",
|
|
"values=tensor([0.3753, 0.2305, 0.2007, 0.2006, 0.1993, 0.1952, 0.1930, 0.1898, 0.1831,\n",
|
|
" 0.1731, 0.1649, 0.1647, 0.1543, 0.1320, 0.1314, 0.1303, 0.1296, 0.1261,\n",
|
|
" 0.1245, 0.1243], grad_fn=<TopkBackward>),\n",
|
|
"indices=tensor([8942, 6336, 1852, 9056, 1865, 4039, 7820, 5002, 8208, 1857, 9709, 803,\n",
|
|
" 1046, 130, 4306, 6481, 4370, 4259, 4285, 1855]))"
|
|
]
|
|
},
|
|
"execution_count": 40,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"torch.topk(list(lr_model.parameters())[0][0], 20)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 41,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"the\n",
|
|
"of\n",
|
|
"christ\n",
|
|
"to\n",
|
|
"church\n",
|
|
"god\n",
|
|
"rutgers\n",
|
|
"jesus\n",
|
|
"sin\n",
|
|
"christians\n",
|
|
"we\n",
|
|
"and\n",
|
|
"athos\n",
|
|
"1993\n",
|
|
"hell\n",
|
|
"our\n",
|
|
"his\n",
|
|
"he\n",
|
|
"heaven\n",
|
|
"christian\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"for i in torch.topk(list(lr_model.parameters())[0][0], 20)[1]:\n",
|
|
" print(vectorizer.get_feature_names()[i])"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 42,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"torch.return_types.topk(\n",
|
|
"values=tensor([-0.3478, -0.2578, -0.2455, -0.2347, -0.2330, -0.2265, -0.2205, -0.2050,\n",
|
|
" -0.2044, -0.1979, -0.1876, -0.1790, -0.1747, -0.1745, -0.1734, -0.1647,\n",
|
|
" -0.1639, -0.1617, -0.1601, -0.1592], grad_fn=<TopkBackward>),\n",
|
|
"indices=tensor([5119, 8096, 5420, 4436, 6194, 1627, 6901, 5946, 9970, 3116, 1036, 9906,\n",
|
|
" 5654, 8329, 7869, 1039, 1991, 4926, 5035, 4925]))"
|
|
]
|
|
},
|
|
"execution_count": 42,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"torch.topk(list(lr_model.parameters())[0][0], 20, largest = False)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 43,
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"keith\n",
|
|
"sgi\n",
|
|
"livesey\n",
|
|
"host\n",
|
|
"nntp\n",
|
|
"caltech\n",
|
|
"posting\n",
|
|
"morality\n",
|
|
"you\n",
|
|
"edu\n",
|
|
"atheism\n",
|
|
"wpd\n",
|
|
"mathew\n",
|
|
"solntze\n",
|
|
"sandvik\n",
|
|
"atheists\n",
|
|
"com\n",
|
|
"islamic\n",
|
|
"jon\n",
|
|
"islam\n"
|
|
]
|
|
}
|
|
],
|
|
"source": [
|
|
"for i in torch.topk(list(lr_model.parameters())[0][0], 20, largest = False)[1]:\n",
|
|
" print(vectorizer.get_feature_names()[i])"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### sieć neuronowa"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 44,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"class NeuralNetworkModel(torch.nn.Module):\n",
|
|
"\n",
|
|
" def __init__(self):\n",
|
|
" super(NeuralNetworkModel, self).__init__()\n",
|
|
" self.fc1 = torch.nn.Linear(FEAUTERES,500)\n",
|
|
" self.fc2 = torch.nn.Linear(500,1)\n",
|
|
"\n",
|
|
" def forward(self, x):\n",
|
|
" x = self.fc1(x)\n",
|
|
" x = torch.relu(x)\n",
|
|
" x = self.fc2(x)\n",
|
|
" x = torch.sigmoid(x)\n",
|
|
" return x"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 45,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"nn_model = NeuralNetworkModel()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 46,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"BATCH_SIZE = 5"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 47,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"criterion = torch.nn.BCELoss()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 48,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"optimizer = torch.optim.SGD(nn_model.parameters(), lr = 0.1)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 49,
|
|
"metadata": {
|
|
"scrolled": true
|
|
},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"0"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.6605833534551934, 0.5908529048207664)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.6379233609747004, 0.6481481481481481)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"1"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.4341224195120214, 0.896168108776267)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.3649017943276299, 0.9074074074074074)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"2"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.18619558424660096, 0.9765142150803461)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.16293201995668588, 0.9888888888888889)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"3"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.09108264647580784, 0.9962917181705809)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.08985773311858927, 0.9962962962962963)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"4"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.053487053708540566, 0.9987639060568603)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
},
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.05794332528279887, 1.0)"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"for epoch in range(5):\n",
|
|
" loss_score = 0\n",
|
|
" acc_score = 0\n",
|
|
" items_total = 0\n",
|
|
" nn_model.train()\n",
|
|
" for i in range(0, Y_train.shape[0], BATCH_SIZE):\n",
|
|
" X = X_train[i:i+BATCH_SIZE]\n",
|
|
" X = torch.tensor(X.astype(np.float32).todense())\n",
|
|
" Y = Y_train[i:i+BATCH_SIZE]\n",
|
|
" Y = torch.tensor(Y.astype(np.float32)).reshape(-1,1)\n",
|
|
" Y_predictions = nn_model(X)\n",
|
|
" acc_score += torch.sum((Y_predictions > 0.5) == Y).item()\n",
|
|
" items_total += Y.shape[0] \n",
|
|
"\n",
|
|
" optimizer.zero_grad()\n",
|
|
" loss = criterion(Y_predictions, Y)\n",
|
|
" loss.backward()\n",
|
|
" optimizer.step()\n",
|
|
"\n",
|
|
"\n",
|
|
" loss_score += loss.item() * Y.shape[0] \n",
|
|
"\n",
|
|
" display(epoch)\n",
|
|
" display(get_loss_acc(nn_model, X_train, Y_train))\n",
|
|
" display(get_loss_acc(nn_model, X_dev, Y_dev))"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 50,
|
|
"metadata": {},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"text/plain": [
|
|
"(0.16834938257537793, 0.9428172942817294)"
|
|
]
|
|
},
|
|
"execution_count": 50,
|
|
"metadata": {},
|
|
"output_type": "execute_result"
|
|
}
|
|
],
|
|
"source": [
|
|
"get_loss_acc(nn_model, X_test, Y_test)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### Zadanie domowe\n",
|
|
"\n",
|
|
"- wybrać jedno z poniższych repozytoriów i je sforkować:\n",
|
|
" - https://git.wmi.amu.edu.pl/kubapok/paranormal-or-skeptic-ISI-public\n",
|
|
" - https://git.wmi.amu.edu.pl/kubapok/sport-text-classification-ball-ISI-public\n",
|
|
"- stworzyć klasyfikator bazujący na prostej sieci neuronowej feed forward w pytorchu (można bazować na tym jupyterze). Zamiast tfidf proszę skorzystać z jakieś reprezentacji gęstej (np. word2vec).\n",
|
|
"- stworzyć predykcje w plikach dev-0/out.tsv oraz test-A/out.tsv\n",
|
|
"- wynik accuracy sprawdzony za pomocą narzędzia geval (patrz poprzednie zadanie) powinien wynosić conajmniej 0.67\n",
|
|
"- proszę umieścić predykcję oraz skrypty generujące (w postaci tekstowej a nie jupyter) w repo, a w MS TEAMS umieścić link do swojego repo\n",
|
|
"termin 25.05, 70 punktów\n"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"author": "Jakub Pokrywka",
|
|
"email": "kubapok@wmi.amu.edu.pl",
|
|
"kernelspec": {
|
|
"display_name": "Python 3",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"lang": "pl",
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.8.3"
|
|
},
|
|
"subtitle": "8.Regresja logistyczna[ćwiczenia]",
|
|
"title": "Ekstrakcja informacji",
|
|
"year": "2021"
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 4
|
|
}
|