The Blog

Immunocomputing: Genetic Algorithms and Databases Inspired by the Human Immune System

This is a small excerpt from Seed Magazine’s article on Fernando Esponda’s work on the Negative Database, designed to work much like the human immune system.

Fernando Esponda, Immunocomputing

Information puts Fernando Esponda in a negative state of mind. Which is exactly why he’s poised to overturn conventional ideas in information science. His innovative research began as he was working toward his doctorate in computer science and became interested in the human immune system. “What caught my eye was that the information being used by the immune system was a negative kind of information,” he says. That is, the immune system doesn’t have a record of every possible pathogen that could invade the body. Instead, it learns what the body itself looks like and knows to go on the offensive when it encounters anything that doesn’t match its definition of “self.” “Can we do the same thing with data?” Esponda began to wonder. “Can we take a database, and instead of storing everything that’s in it, can we store everything that is not in it?”

The idea sounded preposterous, and Esponda’s colleagues told him as much. But in his dissertation, Esponda demonstrated that a negative database could be created, stored, and manipulated effectively and efficiently—setting the stage for a revolution in information science. Now Esponda is working to determine when negative databases might possess significant advantages over the status quo. One obvious application, he says, involves improving data security. Though it is sometimes possible to work backward from a negative database to its positive complement, negative databases are still tougher to crack, especially if they are divided into several parts and stored in separate locations.

Esponda, who works at the Mexico Autonomous Institute of Technology (ITAM), is exploring scenarios in which it could be useful to collect negative data about behaviors that people might find sensitive, and about which they might lie. For instance, he says, researchers can design a survey question asking women how many abortions they’ve had, giving them five answers to choose from. But rather than asking a woman to check the box that applies to her, the survey could ask her to check one of the boxes that does not apply. Reverse surveys reveal only a little about each subject but a lot about a population, and they can accurately estimate how common a behavior is without anyone having to admit to it. “Every choice you make leaves a choice of inaction, of what you did not do,” Esponda says. “I’m trying to show exactly what can be learned from this inaction. How can we gain insight from what is not there?”

Leave a Comment

Let us know your thoughts on this post but remember to place nicely folks!

You must be logged in to post a comment.