Download presentation
Presentation is loading. Please wait.
Published byClaribel Shepherd Modified over 8 years ago
1
Ten MC Questions taken from the Text, slides and described in class presentation. COSC 4426 AJ Boulay Julia Johnson
2
MC Question 1 – p. 92 What simple operation does each neuron in an ANN do? (a) it sums its weighted inputs and applies a certain activation function on the sum. (b) it does a pattern addition. (c) it activates the next neuron. (d) it does a cross product. (answer – (a))
3
MC Question 2 - Presentation What kind of network was the network we trained in class? (a) Hopfield network. (b) Pattern Associator. (c) Self Organizing Map. (d) Restricted Boltzmann Machine. (answer – (a) Pattern Associator)
4
MC Question 3 - p. 93 A Binary Step function is an example of: (a) an activation function. (b) a squashing function. (c) (a) only (d) both (a) and (b).
5
MC Question 4 – p. 93 A sigmoid activation function is also called: (a) the Delta Rule. (b) The Least Mean Square method. (c) (a) only. (d) both (a) and (b).
6
MC Question 5 – p.97 According to the Hebb Rule, how can learning be proved to have occurred? (a)Check the TLEARN software to see if the network has learned. (b) If learning patterns are mutually orthogonal, then it can be proved that learning will occur. C)If learning patterns are related to the output, then it can be proved that learning will occur. D)Learning only occurs with the Delta Rule. E)(Answer B - If learning patterns are mutually orthogonal, then it can be proved that learning will occur. )
7
MC Question 6 – p. 98 What kind of learning is the Delta Rule associated with? (a) Linear Feed Forward Learning. (b) Gradient Decent learning. ( c) Multilayer Learning only. (d) Restricted Boltzmann Machine learning. (Answer (b) – Gradient Decent)
8
MC Question 7 –pp. 99-102 What kind of cases can a Perceptron classify. (a) AND, OR but not Both. (b) Pattern Associated cases. (c) Linearly separable cases. (d) Any kind of binary case. (Answer c – linearly separable cases)
9
MC Question 8 – p 102-103 A Kohonen network is what kind of network? (a) A binary classifier. (b) A Pattern Associator. (c) A fully connected Self Organizing Map. (d) A Restricted Bolzmann Machine. (Answer c – SOM)
10
MC Question 9 – p. 116-122 How is knowledge represented in a Hopfield neural network. (a) Knowledge is in the units. (b) Addressable Content memory. (c) Knowledge is in the weights only. (d) Knowledge Representation. Answer b – Hopfield nets have addressable content memory)
11
MC Question 10 - Presentation What if there is no change in weights in a neural network when you train it? (a) then the network has learned. (b) then the network has not learned. (c) sometimes there is no change in weights when learning. (d) (c) only. (Answer b )
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.