/2021TWBootcamp

Lecture material for Taiwan Summer Bootcamp

Primary LanguageJupyter Notebook

2021TWBootcamp

Lecture material for Taiwan Summer Bootcamp

Installation

Python 3.6 or Anaconda, if you choose to use pip to install package, we recommend using virtualenv to manage your environment.

Below are the steps to setup the required packages for bootcamp's excercises.

Install by pip

pip install -r requirements.txt

Install by conda

conda create --name <env_name> --file requirements.txt python=3.6
conda activate <env_name>

Install PyTorch

PyTorch requires different command based on your environment. In this course, we will use PyTorch 1.9.0 (which is latest at the moment). You choose the command to install by accessing the following link:

https://pytorch.org/get-started/locally/

Syllabus

Date Subject
Aug 9
    Bootcamp Opening Day
  • A Sharing by Hajime
  • Syllabus Introduction (Tsung-Wei)
    		•
    Aug 10
      Lean startup and MVP project
  • Lean Startup +Demo (Paul, Tsungwei)
  • Life Cycle of ML project (Mandy)
    		•
    Aug 12
      Agile Software development
  • SCRUM (Eason)
  • Docker (Jacob)
    		•
    Aug 17
      Git WorkFlow
  • Git (Blake)
      • DataLoader
  • PyTorch (Dini)
    		•
    Aug 19
      CV Course1
  • Image Data Augmentation (Blake)
  • Torchvision.transforms, Imgaug
      • CV Course2
  • Classic CNN Structures (Blake)
  • Advanced CNN Structures - Part1
  • InceptionV1-V4
  • ResNetV1-V2
    		•
    Aug 24
      CV Course3
    Advanced CNN Structures - Part2
  • DenseNet (Jeff)
  • ShuffleNet
  • SENet
  • MobileNetV1-V3
  • EfficientNetV1-V2
  • NFNet
  • GhostNet
      • CV Course4
  • Loss function - Select & Design (Jeff)
    		•
    Aug 26
      CV Course5
    Bag of training tricks (Jeff)
  • Stochastic Depth
  • Warm up
  • Label Smoothing
  • No Bias Weight Decay
  • Teacher-Student Knowledge Distillation
  • Mixup
  • Group Normalization
  • Weight Standardization
      • CV Course6
    Hands On - Build and Serve OCR engine
  • OCR-Captcha-Cracker
    		•
    Aug 31
      NLP Technical Class [1]
    Missions & Metrics (Dini)
  • Tasks of NLP
  • Common Metrics of NLP
      • NLP Technical Class [2]
    Preprocessing (Dini)
  • Removing
  • Normalizing
  • Tokenize
    		•
    Sep 2
      NLP Technical Class [3]
    Basic Encodings and Embeddings (Tsung-Wei)
  • Embedding definition
  • One-hot enocding
  • TF-IDF
  • N-gram
  • Word2Vec
  • Glove
    		•
    Sep 7
      NLP Technical Class [4]
    Language Models (Ian)
  • Language modelling
  • Language models: n-gram, RNN, CNN, Transformer
  • Pretrained models: ELMo, GPT, BERT
    		•
    Sep 9
      CV Course7
    Vision transformer & Evolution (Jeff)
  • DeiT, ViT
  • CaiT
  • CeiT
  • DeepViT
  • VOVO
      • Appendix A
  • Research Review Flow (Dini)
    		•
    Sep 14 Pitching
  • Presentation (Paul)
    		•
    Sep 16 Pre-Rehearsal
    Sep 17 Rehearsal
    Sep -- Demo Day