Improbable Research

The Varshney twins – Dr. Kush Varshney (currently at IBM) and Professor Lav Varshney (previously at IBM) – have authored a series of papers on the theme of computational gastronomy, two of which — on Food Steganography, and on Active Odor Cancellation — we looked at recently.

Example 3 : ‘Flavor Pairing in Medieval European Cuisine: A Study in Cooking with Dirty Data‘ in: International Joint Conference on Artificial Intelligence Workshops, Beijing, China, August 2013. (The paper is co-authored with Jun Wang, also at IBM, and Daniel Myers at Medieval Cookery.) It not only examines the computational gastronomic implications of furmente with purpeys and zedoary, but also eles in brasill and chekin.

“In order to cook with computers, specifically by having the computer use the flavor compound pairing hypothesis to create new sets of ingredients in recipes, we must understand the effect of data dirtiness. Towards that end, in this paper we obtain two independent data sets of flavor compound data having different levels of completeness and accuracy, and compile a completely new corpus of recipes from the Late Middle Ages that has not been analyzed before.”

This concludes our short profile on the Varshney twins’ contribution to the field of computational gastronomics.

BONUS : Co-author Daniel Myers maintains an extensive online list of medieval recipes from across Europe.

If you enjoy fun, conceivably-partly-true tales about disruptive technologies (as some people do — see, for example, Harvard Business School Professor Clay Christensen‘s tidy tales of disruption), then pay heed to the scholarly studies from Russell B. Clayton. Clayton, at the University of Missouri-Columbia, has a new study about Twitter and marital breakups: “The Third Wheel: The Impact of Twitter Use on Relationship Infidelity and Divorce,” Russell B. Clayton, Cyberpsychology, Behavior, and Social Networking, vol. 17, no. 7, July 2014, pp. 425-430….

—so begins another Improbable Innovation nugget, which appears in its entirety on BetaBoston.

What of walks and chocolate eaters? Oh, maybe this:

“Brisk walking reduces ad libitum snacking in regular chocolate eaters during a workplace simulation,” Hwajung Oh [pictured here], Adrian H. Taylor, Appetite, Volume 58, Issue 1, February 2012, Pages 387-392. The authors, at the  University of Exeter, UK, report their discovery:

“A brief walk may help to reduce ad libitum snacking in regular chocolate eaters.”

Ig Nobel Prize winner Mahadevan (2007 Ig Nobel physics prize, for studying how sheets become wrinkled) and colleagues have taken an applied-mathematical look at yet another unanswered question. As happens so often with Mahadevan and his merry, varying band of collaborators, a better-than-anyone-had-before answer appeared. Details are in this study:

“The dynamics of sperm cooperation in a competitive environment,” Heidi S. Fisher, Luca Giomi, Hopi E Hoekstra, L. Mahadevan, BioRXiv, 2014. The authors, at Harvard University and the International School for Advanced Studies, Trieste, Italy, report:

“we use fine-scale imaging and a minimal mathematical model to study sperm aggregation in the rodent genus Peromyscus. We demonstrate that as the number of sperm cells in an aggregate increase, the group moves with more persistent linearity but without increasing speed; this benefit, however, is offset in larger aggregates as the geometry of the group forces sperm to swim against one another. The result is a non-monotonic relationship between aggregate size and average velocity with both a theoretically predicted and empirically observed optimum of 6-7 sperm/aggregate.”

Ed Yong, in his Not Exactly Rocket Science blog, has a nice essay about this.

Here’s further detail from the study itself:

BONUS (possibly not strongly related): The 1956 essay “The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information“

BONUS (probably unrelated): “Researchers find first sign that tyrannosaurs hunted in packs“

The Varshney twins – Dr. Kush Varshney (currently at IBM) and Professor Lav Varshney (previously at IBM) – have authored a series of papers on the theme of computational gastronomy, one of which, on Food Steganography, we looked at recently.

Example 2. Active Odor Cancellation. (IEEE International Workshop on Statistical Signal Processing, Gold Coast, Australia, June-July 2014.)

“Noise cancellation is a traditional problem in statistical signal processing that has not been studied in the olfactory domain for unwanted odors. In this paper, we use the newly discovered olfactory white signal class to formulate optimal active odor cancellation using both nuclear norm-regularized multivariate regression and simultaneous sparsity or group lasso-regularized non-negative regression. As an example, we show the proposed technique on real-world data to cancel the odor of durian, katsuobushi, sauerkraut, and onion.”

Coming soon: Computational gastronomy part 3