day3-recap

123456789101112
Across
  1. 2. Activation function that outputs max(0,x)
  2. 4. Downsampling technique that reduces feature map size
  3. 7. One complete pass through the training dataset
  4. 9. Weight matrix used to extract features in CNN
  5. 10. Mathematical slope used in descent optimization
  6. 11. Distance between consecutive receptive fields
  7. 12. Process of sliding a filter across an image
Down
  1. 1. Regularization technique that randomly disables neurons
  2. 3. Training algorithm that propagates errors backward
  3. 5. When training accuracy exceeds test accuracy significantly
  4. 6. Activation function that outputs probability distribution
  5. 8. Network type where neurons connect to local image regions