MRSEC Education Group
We use materials science to enhance public understanding of STEM concepts and to share the excitement of these fields with K-12 audiences and the general public.

Machine Learning Predicts the Periodic Table

Audience

This activity is for high school students.  This works best after students have learned electron configurations and periodic trends.   

There are few supplies required (1 hand out, white boards, computers and colored pencils.)  

Time Frame

Set-up: 0.5 hours (very minimal set up)

Activity: (4)  45 minute class periods or (2) 90 minute block

Clean-up: 0.1 hours (minimal)

Objective(s)

After completing the activity, participants will be able to:

  1. Students will be able to understand how a machine learns to read, interpret and make predictions using  a data set.
  2. Students will be able find similarities and differences between a computer model of the periodic table and Dmitir Mendeleev’s model.  
  3. Students will learn the limitations of large data sets. 
  4. Students will learn how to interpret and evaluate  data using heat maps.

Standards Addressed 

PS1:   Matter and Its Interactions 

PS1A: Structure and Properties of matter

Engineering principles

Students will work in teams to decipher a computer’s code to the periodic table by breaking it down into recognizable sections.  

HS-ETS1-2. Design a solution to a complex real-world problem by breaking it down into smaller, more manageable problems that can be solved through engineering.

Activity Materials

Safety

None

Activity Instructions

Set Up
  1. Have the students complete Flinn’s “It’s in the Cards Activity” (90 minutes)
    (This is going to mimic human pattern based learning)
  2. Discuss how Mendeleev used specific properties to lay out the current periodic table.
  3. Use the From Code to a Periodic Table google slides to teach the next activity
    1. Slide 2 introduces Machine Learning with 2 videos.  
    2. Slide 3 introduces the students to the data set that the computer used to generate a periodic table.  Using the google form created, have the students analyze what some of the data headings mean.  
    3. Slide 4 reveals the periodic table generated by the computer
      1. Provide the students with the “Decipher the Table” Worksheet in their groups of 2.  
      2. They will work together to find out where various groupings from Mendeleev’s table are on the new table
    4. Slides 5-8 introduce students into a heat map for data presentation.  
      1. Students will work in pairs to analyze the heat patterns for 3 chemical properties.
      2. A class group discussion to compare patterns will follow

Conclusion (15 minutes)

Collect all cards from Flinn Activity.

Have students reflect upon the 2 activities with the 1-4 questions below.

Assessment

Students have a socratic seminar to discuss the pros and cons to machine learning.  

The following statements will be used for students to generate a note sheet for their socratic seminar.  

  1. Were the heat maps easier to see trends on the periodic table?  Why or why not?
  2. Consider the differences in data sets between the “It’s in the Cards” activity to the machine learning model.  What impact on patterns do you think the difference in data sets have?  
  3. After completing this activity, do you believe that a machine is always better at making predictions than humans?  Give 2 specific examples from the activity to support your statement.
  4. What aspects of machine learning would you like to more about?

Socratic Note sheet during Seminar

Background

Students should be able to write electron configurations and speak to periodic trends on the table.  Instructors should be familiar with machine learning in general.  

Supplemental Materials

 

References

Kusaba, M., Liu, C., Koyama, Y. et al. Recreation of the periodic table with an unsupervised machine learning algorithm. Sci Rep 11, 4780 (2021). 

Authors 

Jamie Lauer 

MRSEC IEG Leadership Team: Matthew Stilwell, and Anne Lynn Gillian-Daniel

Principal Investigator:  Jacob Notbohm

Mentors: Ben Afflerback, Kelly Vasquez and Christa Wille