CS486 - Introduction to Neural Networks and Deep Learning

123456789101112131415161718192021222324252627282930
Across
  1. 5. ___________ in the biological neuron is referred to as output in computational neuron.
  2. 6. Weights in the Rosenblatt's perceptron model are __________________ numbers
  3. 9. A NN model with 1 - Input layer, 4 hidden layer and 1 Output layer is referred to be ______ layer network
  4. 10. __________ layer can be used to avoid overfitting
  5. 13. No of Convolution layers used in VGG16 is
  6. 14. a Layer that reduces the dimensionality of the convoluted image
  7. 19. a popular algorithm used to minimize the loss function
  8. 20. To make the convoluted image the same size as the input image, then the padding value should be ‘________’ in Python.
  9. 21. Use the _______________ of pre-trained models when the dataset is huge and similarity is less.
  10. 23. a constant used to update the net activation value
  11. 25. A number 26 is an example of a _____________
  12. 26. ____________ learning mechanism is used when the dataset is small and similarity of images is less with Imagenet while using pre-trained models
  13. 27. An input layer has 'm' units and hidden layer has 'n' units then the number of weights between these two layers is ___________
  14. 28. ___________ is an example of linearly inseperable problem
  15. 29. A grayscale image is an example of _______________
  16. 30. In the McCulloch Pitts' model, the inputs are
Down
  1. 1. an activation function used to predict the output in logistic regression
  2. 2. one ____________ refers to show the dataset completely to the NN model.
  3. 3. ________________ Loss function is used to update the weights during multiclass classification.
  4. 4. pre-trained models are used as ___________ when the dataset is huge and similarity is also high.
  5. 7. a layer that produces probabilities of a neuron in a classification
  6. 8. an alternate name for threshold activation function
  7. 11. One _____________ refers to show the one instance of the dataset to the NN model.
  8. 12. a hyperparameter used to determine how fast the model can learn from data
  9. 15. If a CNN model provides Test error = 0.22 and Train error = 0.01, then the model is an _______________ model
  10. 16. a popular loss function used in adjusting the weights of a neural network
  11. 17. __________ activation function is used in regression process.
  12. 18. Gradient descent refer to moving towards the ______________ slope of the gradient
  13. 22. a non-linear activation function that squashes the output in -1 to +1
  14. 24. Activation function used to predict the binary output in the final dense layer is _________