Similar to a mouse running through a maze, making ‘yes’ or ‘no’ decisions at every intersection, researchers have developed a way for machines to quickly learn all the twists and turns of a complex data system.
“Our method can help improve diagnosis of urinary disease, imaging of heart disease, and analysis of financial risk,” reported Abd-AlRahman Rasheed AlMomani of the Embry-Riddle Aeronautical University campus in Prescott, Arizona. .
The research has been accepted for the Nov. 11 edition of the peer-reviewed journal Patterns, an imprint of Cell Press, with Jie Sun and Erik Bollt of Clarkson University’s Center for Complex Systems Science. The goal of the work is to more efficiently parse binary (“Boolean”) data.
“We can see everything around us as a web of objects and variables interacting with each other,” said AlMomani, assistant professor of data science and mathematics at Embry-Riddle. “Understanding these interactions can improve our predictions and the management of a whole range of networks – from the regulatory networks of biology and genes, to aerial flight.”
Boolean or “yes/no” data is frequently used in the field of genetics, where gene states can be described as “on” (with high gene expression) or “off” (with little or no gene expression). gene), AlMomani explained. Learning Boolean functions and networks based on noisy observation data is essential for deciphering many different scientific and engineering problems – from pollinator plant dynamics and drug targeting to tuberculosis risk assessment of a person.
The challenge, AlMomani explained, is that the standard method of learning Boolean networks — called REVEAL (for Reverse Engineering Algorithm for Interference of Genetic Network Architectures) — mixes many different sources of information. The REVEAL approach thus increases computational complexity and cost, and researchers need to dampen the noise to analyze all the data. Moreover, the REVEAL method is not optimal for solving quantitative biology problems, which require discovering causal factors.
To eliminate incorrect answers more quickly, AlMomani and his colleagues used a method called Boolean Optimal Causal Entropy, which gradually reduces the number of correct solutions to a problem. The method essentially transforms a complex diagnostic process into a decision tree, where yes/no questions such as “Does the patient have a fever?” Nausea? Lower back pain?” can guide a clinician to the correct diagnosis.
AlMomani explained that many different scientific questions depend “on a Boolean variable which is basically zero or one. An event has occurred or it has not occurred. A patient will take a test and get a positive or negative result. We can then categorize that patient’s test results, medical history, and results as Boolean variables. »
To test their ideas, the researchers got their hands on a full set of 958 possible board configurations at the end of a game of Tic-Tac-Toe. The board and the different moves of the game were then expressed as mathematical problems in order to predict which player would win.
The researchers also tested their method using a dataset from cardiac spectroscopy images. Their system got the correct diagnosis 80% of the time.
The Patterns paper, “Data-Driving Learning of Boolean Networks and Functions by Optimal Causation Entropy Principle (BoCSE)”, was funded in part by the U.S. Army Research Office (Grant W911NF-16-1-0081) and the Simons Foundation (grant 318812).
DOI: 10.1016/j.patter.2022.100631
After the embargo has lifted, this document is available online at https://www.cell.com/patterns/fulltext/S2666-3899(22)00263-X.
About Embry-Riddle Aeronautical University
Journalists around the world contact Embry-Riddle Aeronautical University for content experts in all aspects of aviation, aviation business, aerospace, engineering, and STEM-related fields. Our faculty experts specialize in unmanned and autonomous systems, security and intelligence, air traffic and airport management, astronomy, human factors psychology, meteorology, spaceflight operations, urban air mobility and much more. Visit the Embry-Riddle Newsroom for story ideas.
Embry-Riddle educates more than 32,500 students at its residential campuses in Daytona Beach, Florida and Prescott, Arizona, at approximately 110 campuses worldwide and through online degree programs. U.S. News & World Report named Embry-Riddle Worldwide as the top national provider of online licensing programs.