paranormal-or-skeptic-ISI-p.../regression.py
2021-05-25 23:57:06 +02:00

85 lines
3.2 KiB
Python

import numpy as np
import pandas as pd
import torch
from csv import QUOTE_NONE
from nltk.tokenize import word_tokenize
import gensim.downloader
#Based on source material from classes
class MyNeuralNetwork(torch.nn.Module):
def __init__(self, input_size, hidden_size, num_classes):
super(MyNeuralNetwork, self).__init__()
self.fc1 = torch.nn.Linear(input_size, hidden_size)
self.fc2 = torch.nn.Linear(hidden_size, num_classes)
def forward(self, x):
x = self.fc1(x)
x = torch.relu(x)
x = self.fc2(x)
x = torch.sigmoid(x)
return x
word2vec = gensim.downloader.load('word2vec-google-news-300')
def get_word2vec(document):
return np.mean([word2vec[token] for token in document if token in word2vec] or [np.zeros(300)], axis=0)
#Basic paths + reading from files
XtrainingData = pd.read_table('train/in.tsv.xz', error_bad_lines=False, header=None, quoting=QUOTE_NONE, names=['content', 'id'])
YtrainingData = pd.read_table('train/expected.tsv', error_bad_lines=False, header=None, quoting=QUOTE_NONE, names=['label'])['label']
XtestData = pd.read_table('test-A/in.tsv.xz', error_bad_lines=False, header=None, quoting=QUOTE_NONE, names=['content', 'id'])
XdevData = pd.read_table('dev-0/in.tsv.xz', error_bad_lines=False, header=None, quoting=QUOTE_NONE, names=['content', 'id'])
#Data filltering and preprocessing
XtrainingData = [word_tokenize(row) for row in XtrainingData['content'].str.lower()]
XtestData = [word_tokenize(row) for row in XtestData['content'].str.lower()]
XdevData = [word_tokenize(row) for row in XdevData['content'].str.lower()]
XtrainingData = [get_word2vec(document) for document in XtrainingData]
XtestData = [get_word2vec(document) for document in XtestData]
XdevData = [get_word2vec(document) for document in XdevData]
#Basic parameters for the model
eph = 30
batches = 5
network = MyNeuralNetwork(300, 600, 1)
criterion = torch.nn.BCELoss()
optimizer = torch.optim.SGD(network.parameters(), lr=0.02)
#Model training according to source files from classes
for epoch in range(eph):
network.train()
for i in range(0, YtrainingData.shape[0], batches):
x = XtrainingData[i :i + batches]
x = torch.tensor(x)
y = YtrainingData[i :i + batches]
y = torch.tensor(y.astype(np.float32).to_numpy()).reshape(-1, 1)
outputs = network(x.float())
loss = criterion(outputs, y)
optimizer.zero_grad()
loss.backward()
optimizer.step()
#Basic evaluation
YpredDev = []
YtestPred = []
with torch.no_grad():
for i in range(0, len(XdevData), batches):
x = XdevData[i :i + batches]
x = torch.tensor(x)
outputs = network(x.float())
prediction = outputs > 0.5
YpredDev += prediction.tolist()
for i in range(0, len(XtestData), batches):
x = XtestData[i :i + batches]
x = torch.tensor(x)
outputs = network(x.float())
prediction = outputs > 0.5
YtestPred += prediction.tolist()
#Saving outputs
np.asarray(YpredDev, dtype=np.int32).tofile('./dev-0/out.tsv', sep='\n')
np.asarray(YtestPred, dtype=np.int32).tofile('./test-A/out.tsv', sep='\n')