Q1: What is the most significant difference between Hopfield Neural Network and Convolutional
Neural Network - as studied in this course (Textbook Chapters 1-9)?
Q2: (a) What is a Restricted Boltzmann Machine, and (b) what is it successfully used for?
Q3: (a) What is a Deep Learning Neural Network, (b) why and how does it work?
Q4: What are three important deficiencies of Neural Networks with Linear InnerProduct
Kernel(s)?(list each deficiency separately, and why/how it occurs)
Q5: (a) What does this Diagram mean (discuss type of NN, all features & function)? (b) Why do
the boxes get smaller in two dimensions but sometimes get thicker along the propagation axis,
and (c) what advantage does this provide in a successful implementation?
Q6: Given loss function L, weights hidden units , and output o, what does this
equation mean for Backpropagation? Explain in your own words the MEANING of EACH and
ALL operations & variables, and WHY they are used - DO NOT COPY FROM ANY SOURCE:
Q7: Discussion: Compare and Contrast Perceptron NNs, Radial Basis Function NNs, and
Lattice-based NNs in terms of (a) basic mathematical functionality, (b) primary application(s),
and (c) key deficiencies. Explain in your own words how each of these features or differences
arise, and why each isimportant: DO NOT COPY FROM ANY SOURCE (3 sentences per part)
Q8:(a) What are Lattice-Based Neural Networks (LNNs), and (b) why are LNNs with dendritic
computation potentially more noise-tolerant than classical NNs such as Feed-Forward
Networks? (Explain in salient detail and you may use a simple diagram for b)