cs231n mindmap

a brief overview of Stanford CS231n, "Deep Learning-based Computer Vision"

CS231n deep learning

• tasks
  • classification
    • space
      • image
        • kNN
        • linear classifier
      • feature
    • loss
      • objective
        • hinge loss
        • softmax loss
      • regularization
        • l2 loss
        • data augmentation
    • optimization
      • back propagation
      • methods
        • SGD
        • Adam
  • segmentation
    • semantic
    • instance
  • object detection
  • captioning
  • image generation
    • VAE
    • GAN
• models
  • fully-connected
  • CNN
    • layers
      • convolution
      • relu
      • pooling
      • batch normalization
      • dropout
    • visualization and understanding
      • saliency map
        • occlusion
        • backpropagation
      • image reconstruction from features
        • guided backpropagation
        • gradient ascent
      • adversarial examples
        • deepdream
      • neural style transfer
        • content
        • style
          • Gram matrix
  • RNN
    • LSTM
    • GRU