Cogs 2010 Week 2 Questions (due at the end of the lab session)

 

Prior to the session: Read CMC Intro and Chs 1 and 2, at http://www.csee.uq.edu.au/~cogs2010/cmc/

At the lab: work through CMC Chs 3 and 4. Only answers to the questions below need to be handed in.

 

Q1. Describe the function, and how to create each of the following components of a BrainWave network:

            Unit, weight, label, graph

           

Q2. Explain what happens each time you click on the cycle button.

 

Q3. Explain how to create a graph that tracks the activation of a unit.

 

Q4. Describe the architecture of the Jets and Sharks network in terms of the

            Number of units

            Number of weights

            Arrangement of units

 

Q5. Explain how Lance’s properties can be retrieved.

 

Q6. Explain how the prototype Jet can be activated in the network.

 

Q7.  Generalisation Question: Consider an interactive activation and competition (IAC) network that represents a glass of orange juice or cup of coffee, tea or hot chocolate.  The network has pools of features for {OJ, coffee, tea or hot chocolate}, {milk, lemon, neither}, {sugar, honey, no sugar}, and a pool for people’s names {Xanthe and Yummi}.  You don’t need to simulate this network using Brainwave, just do a paper and pencil design and answer the following questions.

  1. Draw an IAC network that will represent each of the pools, and all the inhibitory links.
    [Hint: The network will need five pools of units: one for names, one for instance units, and three for features. Each pool will have 2, 3, or 4 units]
  2. How many units do you need for each pool of the network?
  3. How many negative weights are needed in each pool?
  4. Draw in the positive connections required to store the patterns for (1)  Xanthe’s coffee {coffee, black, no sugar} and (2) Yummi’s tea {tea, lemon, honey}.
  5. How many positive weights are needed to store these two patterns?
  6. Suppose you collected preferences for drinks from 100 people and added them in to the network. How would you use the resulting model to find out how your sample of people typically take their tea?