Applied Problems of Deep Learning (APDL)

• Lecture and seminar materials for each week are in ./week* folders, see README.md for materials and instructions
• Any technical issues, ideas, bugs in course materials, contribution ideas - add an issue

Syllabus

• week01 Introduction to Digital Signal Processing

  • Lecture: Signals, Fourier transform, Spectrograms, MelScale
  • Seminar: Intro in PyTorch, AudioMNIST Classification

• week02 Automatic Speech Recognition

  • Lecture: Metrics, Attention, LAS, CTC, BeamSearch
  • Seminar: QuartzNet and CTC based ASR

• week03 Text to Speech

  • Lecture: Tacotron2, WaveNet, Parallel WaveGAN
  • Seminar: WaveNet based Vocoder

• week04 Language Modeling

  • Lecture: Word Embeddings, Language Modeling
  • Seminar: POS tagging based on bi-LSTM, Text Classification based on CNN+Attention, and Neural LMs

• week05 Machine Translation

  • Lecture: Encoder-Decoder framework, Attention, Transformer
  • Seminar: Transformer in details

• week06 Transfer Learning

  • Lecture: ELMo, BERT
  • Seminar: Intro to HuggingFace

• week07 Facial Recognition

  • Lecture: Triplet loss, Angular Softmax, ArcFace
  • Seminar: Metric Learning with CIFAR100

• week08 Segmentation

  • Lecture: Upsampling, U-Net, HRNet, Metrics
  • Seminar: Cell Segmentation

• week09 Object Detection

  • Lecture: R-CNN, Fast R-CNN, Faster R-CNN, YOLO
  • Seminar: People Detection

• week10 How to deploy your neural network?

  • Lecture: Quantization, Pruning, Distilation
  • Seminar: Flask and torchserve for model deployment

Homeworks

• ASR Implementation of a small ASR model based on QuartzNet
• MT Implementation of a small MT model based on Transformer
• CV Implementation of a small segmentation model

Contributors & course staff

Course materials and teaching (mainly) performed by

• Alexander Markovich