Consider a neural language model (NLM) that uses a softmax layer as its output layer. How many nodes where there be in the output layer? What is the typical interpretation of the value of each output node

Consider the back propagation neural network as shown in figure 1, assume that the neurons have a logistic sigmoid activation function, do the following: a) Perform a forward pass on the network. b) Perform a reverse pass (training) once (target =1, x=1, n = 0.9) 2 0.004 0.01 0.101 0.35 Z1 0.66 0.205 h 0.703 0.404 0.595 0.55 0855 0.9 0.5 0.072 h 0.246 Z2 0.567 0.445 Figure 1 0.057 0.343 0.007

Suppose we are fitting a neural network with three hidden layers to a training set. It is found that the cross validation error Jcv(0) is much larger than the training error Jtrain (0). Should we increase the number of hidden layers?

Consider a block stacking robot with the following actions: 1 Stack(x, y) Preconditions: Clear(y), Holding(x) • Effects: armEmpty, On(x, y), ¬Clear(y), ¬Holding(x) . Unstack(x, y) • Preconditions: Clear(x), On(x, y), armEmpty • Effects: ¬armEmpty, ¬On(x, y), Clear(y), Holding(x) 1 Pickup(x) Preconditions: Clear(x), On(x, TABLE), armEmpty • Effects: ¬armEmpty, ¬On(x, TABLE), Holding(x) • Putdown(x) • Preconditions: Holding(x) • Effects: armEmpty, On(x, TABLE), ¬Holding(x) Create a plan for each of the initial state/goal pairs below Assume armEmpty is in initial state and the table has infinite space А Initial state: B A Goal state: B a. A В Initial state: В Goal state: A b. A C В Initial state: A В Goal state: C C.