This repository contains three notebooks that explore different applications of neural networks, including traditional neural networks, convolutional neural networks (CNNs), and recurrent neural networks (RNNs). Below are the details of the included notebooks:
-
Neural Networks for Image Classification
This notebook uses a neural network to classify images of clothing items using the Fashion MNIST dataset. The model is implemented with Keras and TensorFlow. Fashion MNIST is a dataset with images of clothing and presents a more interesting challenge than the traditional MNIST dataset. More information about the dataset is available here. -
Convolutional Neural Networks (CNNs) for Image Classification
This notebook works with convolutional neural networks (CNNs) to classify images of characters from the TV show The Simpsons. The dataset includes images extracted directly from episodes of the series. You can download the training dataset here (~500MB) and the test dataset here (~10MB). -
Recurrent Neural Networks (RNNs) for Time Series Forecasting
This notebook explores the use of recurrent neural networks (RNNs) to predict the behavior of time series. A temperature dataset is used to forecast future values of the time series. While this task is traditionally performed with ARIMA models, the notebook demonstrates how RNNs can be an effective alternative for time series modeling.
Each notebook is designed to provide practical insights into neural networks in various contexts and problems. Explore the notebooks to see how these models are implemented and how you can apply them to your own data!