Online version of Neural Networks Activity

Resource type

Use type

Tools

Project name

Activity Overview

An online version of the Neural Network Activity, where students simulate the working of a neural network by choosing 4 words to describe an image, which are fed forward through the network, and the results evaluated.  The input weights are adjusted (back propagation) and the process is repeated with a second image. 

Total Activity Time:  2 days, 20 minutes each day

Materials:

Students need access to

  • online activity worksheet (group access)
  • Image 1 and Image 2 for the input layer nodes (revealed only to students playing the input layer nodes)
  • Exit Ticket (google form)

 

Options for Differentiation

  • Assign students to roles / nodes in the Neural Network based on ability. Input nodes need to view and image and describe what they see in just 4 words. Creative thinkers who enjoy writing might be well suited. Hidden nodes receive 4 words and need to choose 2 to pass on. This is one of the more simple tasks, involving a quick decision. Students who might be challenged by the other two roles may find the influence they have over the network in this role engaging. There is only one output node in this model. This role involves thinking and reasoning, ideally, aloud. Then crafting a sentence from the given words. Thinkers with a strong verbal and writing skill set might enjoy this role. Students who enjoy quantification and identifying patterns might enjoy an alternative role / responsibility of leading the class in their development of an evaluation method, How do we know how well we did?
  • Split the class into groups (according to the number of teachers available to be in each group). Design one of these groups to be a relatively smaller group to work closely with a teacher as they complete the activity together. This teacher can plan on reading all directions and walking the students each step of all three activities (Make sure this teacher is available for all three lessons).
 
Log in or register to view attachments and related links, and/or join the discussion