Convolutional Neural Network

Edit Answers
12345678910
Across
  1. 4. - regularization for feature selection
  2. 8. - small matrix used to detect patterns
  3. 9. - Updates weights using gradients
  4. 10. - Avoid vanishing gradient
Down
  1. 1. - multi-scale feature extraction
  2. 2. - Uses small 3×3 filters and deep architecture
  3. 3. - alternate name for depth of the feature map
  4. 5. - Converts outputs into probabilities
  5. 6. - Converts multidimensional data into 1D for dense layers.
  6. 7. - Randomly disables neurons during training to prevent overfitting.