Lightweight Artificial Intelligence Architecture Inspired by Bacterial Gene Networks Being Developed
With funding from the National Science Foundation, a new model is being researched that will adapt the energy efficiency and adaptation capabilities of bacteria to artificial intelligence.
School of Computing Associate Professor Sasitharan Balasubramaniam has received a grant of $782,358 from the National Science Foundation to develop a lightweight, energy-efficient artificial intelligence architecture inspired by bacterial gene regulatory networks.
Bacteria-Inspired Artificial Intelligence
While traditional artificial intelligence systems are inspired by brains and neural networks, this project examines bacteria, which have the ability to adapt to environmental conditions even without brains.
Bacterial gene regulatory networks act as control systems that turn genes on and off in response to internal and external signals, offering energy efficiency.
Artificial Non-Neuronal Network Model
The research team aims to develop a new architecture called artificial non-neuronal networks by examining gene regulation processes.
This approach paves the way for compact computing systems by offering a lower margin of error with far fewer parameters compared to traditional models.
Research and Testing Process
Within the scope of the study, information processing and memory mechanisms in bacterial networks will first be examined, followed by the creation of mathematical models.
The developed software tools will be tested on field-programmable gate arrays and evaluated in terms of performance, scalability, and memory usage.
Application Areas and Expectations
This architecture, which consumes less energy, will expand application areas in miniature devices such as sensors, wearable technologies, and remote monitoring systems.
In this way, devices will be able to stay in environments such as inside the body or soil for longer periods, increasing their data collection capacity.