Research Paper Puts Sensory Evaluation in the Higher Ed Classroom

Armonía (Harmony) by Remedios Varo (1956)

A peer-reviewed paper I co-authored with collaborators at the University of Michigan is published in the current edition of the Journal of Statistics and Data Science in Education (JSDSE). The paper outlines a pilot educational module at the University of Michigan that integrates machine olfaction into an undergraduate deep learning course. By combining chemistry, machine learning, and sensory evaluation, the curriculum challenges students to work with subjective sensory data, demonstrating the feasibility of teaching multimodal data science.

The inclusion of sensory evaluation in a deep learning course is novel as far as papers accepted by JSDSE go. The good news is that our pedagogy-driven endeavor doesn't end with our paper. Dr. Ambuj Tewari received a second New Initiatives/New Instructions (NINI) grant from the University of Michigan that will allow us to take our findings to the next level (which is why I'll be busy teaching this summer). 

Curious? Read the paper and sniff out the facts in the sensory evaluation portion. Readers of Glass Petal Smoke who work in the flavor and fragrance industry will appreciate it. Perfume fans may also find it interesting, and don't have to worry about decoding a research paper. It's all about learning in small incremental steps. The more you do it, the better you get at it.

Citation: Han, Y., Kydd, M. K., Ward, J., & Tewari, A. (2026). Teaching Machine Olfaction in an Undergraduate Deep Learning Course: A Pilot Integrating Chemistry, Machine Learning, and Sensory Evaluation. Journal of Statistics and Data Science Education, 34(3), 372–380. https://doi.org/10.1080/26939169.2026.2665103

Notes & Curiosities

Teaching 100+ students in a large auditorium how to evaluate smells using fragrance blotters was and remains one of the most satisfying teaching experiences I've had at the University of Michigan to date. It proved what I've always known by experience; that sensory evaluation is a powerful multisensory teaching and learning tool when combined with other disciplines. The sense of smell is memory's sense; it doesn't get more pedagogical than that.

I have many students to thank with regard to the evolution of Smell & Tell lectures, especially those that were part of summer boot camps held by MSTEM Academies at the university. Thanks also go out to students who invited me to give a TEDxUofM talk in 2015 (a talk that resonated with a wider audience after the COVID-19 pandemic) and students who shared their anosmia and congenital anosmia stories with me after the event. 

BTW: One of the reasons why Dr. Tewari and I ended up collaborating was the presence of my TEDxUofM talk online; the rest, as they say, is history.

I would be remiss if I didn't thank the Ann Arbor District Library for saying yes to monthly "Smell & Tell" events in 2012, and every year afterward. There isn't a day that goes by that I don’t forget how much creative freedom the library gives me. Honestly, writing this post makes me feel like I'm dreaming with my eyes open.  

Smell & Tell Event | AI, Machine Learning and Smells!

Spock smells "the spores" and sees the future.

Smell & Tell | AI, Machine Learning and Smells! 

Date: Wednesday, September 20, 2023 

Time: 5:30PM-7:30PM 

Location: Ann Arbor District Library (Downtown) 

Address: 343 S 5th Ave, Ann Arbor, MI 48104 

Phone: 734-327-4200

You really need to think like a Vulcan when it comes to prognostication and AI. The future is unknown and as such is subject to fantasy, ideology and cow pies. This is especially true of the intersection of smell, machine learning, and artificial intelligence. 

Machines that analyze smells aren't new. The pairing of Gas Chromatography with Mass Spectrometry (GC-MS) was demonstrated in 1955-56 by Dow scientists in Midland, Michigan. The technology, which is used to analyze individual components of complex mixtures in analytical chemistry labs, continues to evolve. 

We'll explore eclectic smell terrain and collectively follow our nose to get a better understanding of how humans evaluate smells against the capacity of machines that analyze complex mixtures and their components. 

The Smell and Tell series of art+science programming is led by Michelle Krell Kydd, a trained nose in flavors and fragrance who shares her passion for gastronomy, sensory evaluation and the perfume arts on Glass Petal Smoke. Smell & Tell builds community through interactions with flavor, fragrance and storytelling. 

Notes:

Nose-forward reading for this Smell & Tell program:

More than meets the AI: The possibilities and limits of machine learning in olfaction by Ann Sophie Barwich and Elizabeth A. Lloyd (2022)

Gas Chromatography - Mass Spectrometry via Dow, Inc. and the American Chemical Society.