Machine learning playground

• Collection of projects used for learning new concepts.
• The models are implemented in PyTorch.

Structure

Computer vision

• Image classification
  • Doing binary/multi-class classification.
• Models
  • Implementing SOTA convolutional neural networks.
• Object detection
  • Implementing YOLO, SSD, RetinaNet and Faster R-CNN.
• Semantic segmentation
  • Implementing FCN-8 and U-Net.
• Neural style transfer
  • Implementing a neural algorithm for artistic style.
• Generative models
  • Implementing DCGAN, Wasserstein GAN, VAE, Conditional VAE and VQ-VAE.

Natural language processing

• Sentiment analysis
  • Doing sentiment analysis using different pre-trained methods: word-embeddings, BERT.
• Text synthesis
  • Synthesis text (character level).
• word2vec
  • Implementing word2vec.
• Speech classification
  • Classifying speech commands.
• Tokenizer
  • Implementing BPE tokenizer.

Additional

• Collaborative filtering
  • Implementing collaborative filtering.
• Entity embedding
  • Implementing entity embeddings.
• Quantization
  • Quantize a cnn model.
• Pruning
  • Pruning a cnn model.
• Transformers
  • Implementing a Transformer from scratch.