This repository contains a collection of my experiments with deep learning using Keras and TensorFlow. The focus is on exploring various neural network architectures and techniques for tasks like:
- Recurrent Neural Networks (RNNs):
- rnn directory: Different RNN architectures (LSTM, GRU) for tasks like text generation and mood classification. This directory also includes experiments with word embedding techniques to represent text data as dense vectors
- Datasets:
- train_data.txt: Contains a collection of positive sentences.
- train_data_bad.txt: Contains a collection of negative sentences.
- Image Colorization:
- colorization.py: Experimenting with convolutional neural networks to colorize grayscale images.
- Style Transfer:
- styletransfer.py: Implementing neural style transfer to apply the artistic style of one image to another.
- Generative Adversarial Networks (GANs):
- gan_with_vae.py: Implementing a GAN with Variational Autoencoder (VAE).
- Dropout and Batch Normalization:
- This repository also includes experiments exploring the effects of dropout and batch normalization techniques on model performance.
Getting Started:
- Clone this repository:
git clone https://github.com/mateusxap/Keras-Tensorflow.git- Install the necessary libraries:
pip install tensorflow keras- Explore the different scripts and run the experiments.
Disclaimer:
This repository is meant for personal exploration and learning. The code is provided as-is, and some experiments might be in early stages of development.