/Neural-Network-Interviews

Neural Network Interviews Lesson Plan

title layout weight hidden
Neural Network Lesson Plan v0.2
post
10
false

Neural Network Interviews

Course: DS
Mod: 4(?)
Topic: Neural Networks
Amount of time: 1.25 hours
Author: David Braslow

Lesson Summary:

Topic:

  • Neural Networks
  • Multi-Layer Perceptrons

Learn.co material:

(needs updating for Curriculum V2.0)

Prerequisite knowledge/ Prework:

Prerequisite knowledge: The prework activity should be assigned before/while students work on the Learn.co material above. The in-person lesson should be scheduled for a date after students have worked through that material.

Prework: Instruct class to do the following writing exercise and send it to the instructor at least two hours before class. You can send out the file "Neural Network Interview Questions Notecatcher.docx" along with the instructions. Here is some text you can use (fill in bold fields):

In class on **DATE**, we will do an simulation exercise where you will respond to technical interview questions about neural networks. This will be great practice for when you are really interviewing, and it will help you deepen your understanding of the material. In preparation for class, please write one-paragraph answers to the questions below:

1. What is a neural network, and why is it called that?
2. How do neural networks work?
3. What are some key factors to consider when designing a neural network?
4. What are the advantages of neural networks?
5. What are the disadvantages of neural networks?

Your audience for these paragraphs is an interviewer with technical data science knowledge. The class activity will involve refining our answers collaboratively, so please complete this assignment independently. I recommend that you work on them as you progress through the Learn.co notebooks as a way to solidify you understanding of the material. You can use what you've learned from Learn.co to answer these questions, as well as any other resources you find helpful online (and please share any good ones with the class!). Please send INSTRUCTOR AND COACH your answers by DEADLINE. This is not meant to be stressful: your answers don't need to perfect, and if you start to worry that you might have trouble meeting this deadline, please do not hesitate to reach out to us. I have attached a Word doc where you can write your answers and which you may want to use during the class activity.

I suggest sending this out before students start working on the Neural Network topics, as having a prompt that they know they'll have to answer will probably help them get more out of the material. This is because they will process the material with an eye toward using it in the future, rather than passively reading through it. You will likely need to send multiple reminders between when you send it out and the deadline.

Once the deadline has passed, you will need to follow up with students who haven't responded to see what is holding them up, providing support as needed. If any students are totally lost, work to get them to the point of understanding to be able to respond to the first question.

You will also need to skim students' answers before class to ensure that they have completed the assignment and to determine which groups to assign students to. Assign students to five groups corresponding to the five questions. Students should be assigned to groups corresponding to questions that they at least tried to answer, although the answer does not need to be strong. You will also need to divide each of those groups up into subgroups of 2 or 3. I recommend forming groups and subgroups of mixed ability, so students who may have struggled with the activity can have someone to learn from.

In the slide deck, there are visual templates for the group assignments that you will need to fill in. The activity asks students to move between groups in complex ways, which can be confusing. Please fill the templates in accordingly:

Each student will have a code for the lesson, such as 1B or 4A. This code will appear in three places in the group charts on the slide: once for pairs, once for single-question groups, and once for jigsaw groups. Assign a code to each student, and for each student, fill that student's name in everywhere you see their code. Some spaces will likely be blank, depending on the size of your class. To the greatest extent possible, have students in pairs for the first activity.

Learning goals for this lesson:

  1. Explain what neural networks are, how they work, and their properties
  2. Practice responding to technical interview questions about neural networks

Misconceptions:

See "Notes About Today's Activity" Slide:

Materials

  • Slide deck with student names (see Prework)

Lesson Outline:

The PowerPoint slides provide the structure for this lesson. Use these slides to structure the lesson, and follow the timings on Slide 5. Notes on each slide provide additional presentation tips.

Below is an overview of the lesson as structured in the slides

Step: Problem
Time: Slides 1-6: 10 min

Goal/Scenario:
Convince your mock interviewer that you have a deep understanding of neural networks.

Learning Goals in sequence:

  1. Explain what neural networks are, how they work, and their properties
  2. Practice responding to technical interview questions about neural networks

Step: Activation
Time: 0 min

See Prework. Task 1 further serves to activate students' memories about what they learned/wrote.

Step: Learning Goals 1 and 2 and 3, Integration and Assessment:
Time: Pair work: 8 min Group work: 15 min Jigsaw groups: 20 min Interview practice: 15 min Total: 58 min

Students work in pairs/trios to refine one answer to one question. They then join up with another pair/trio to refine answers further, becoming "expert" in that question. Next they jigsaw, mixing between groups so at least one "expert" is in each group. In this group they share their expertise by giving their answers and answering follow-up questions. Lastly, they return to their original partner/trio to practice doing the full interview.

Some suggestions for how to handle students who are stuck:

  • During pair work, work with students whose submitted answers were weak to help deepen their understanding and/or strengthen their explanations
  • During group work, circulate and intervene if you hear weak/incorrect explanations
  • During jigsaws and interviews, give positive reinforcement to encourage struggling students to continue, such as by pointing out strengths in their explanations

Groups share out recommendations and supporting findings. Informal assessment throughout group work.

Step: Reflection & Assessment:
Time: 7 min

Assessment and reflection using questions on Reflection slide.