Zoologists from the University of Cologne studied the nervous system of insects to investigate principles of biological brain computation and possible effects on machine learning and artificial intelligence. In particular, they analyzed how insects learn to associate sensory information in their environment with a food reward, and how they can later retrieve this information to solve complex tasks such as finding food. The results suggest that the transformation of sensory information into memories in the brain may inspire future machine learning and artificial intelligence applications to solve complex tasks. The study was published in the journal PNAS.
Living organisms show remarkable abilities in dealing with problems that arise from complex and dynamic environments. They can generalize their experiences to quickly adapt their behavior as the environment changes. The zoologists investigated how the fruit fly’s nervous system controls its foraging behavior. Using a computer model, they simulated and analyzed the calculations in the fruit fly’s nervous system in response to odors emanating from the food source. “We initially trained our fly brain model in the same way as we trained insects in experiments. In the simulation, we presented a certain fragrance together with a reward and a second fragrance without a reward. After just a few fragrance presentations, the model quickly learns a robust representation of the rewarded fragrance and can then find the source of this fragrance in a spatially complex and temporally dynamic environment, ”said the computer scientist Dr. Hannes Rapp, who created the model as part of his doctoral thesis at the UoC’s Institute of Zoology.
The created model is thus able to generalize from its memory and apply what has been previously learned in a completely new and complex olfactory molecule landscape, while only a very small database of training samples is required for learning. “For our model we use the special properties of biological information processing in the nervous system,” explained Professor Dr. Martin Nawrot, lead author of the study. “In particular, this is the rapid and parallel processing of sensory stimuli through short nerve impulses and the formation of a distributed memory through the simultaneous modification of many synaptic contacts during the learning process.” The theoretical principles underlying this model can also be used for artificial intelligence and autonomous systems. They allow an artificial agent to learn much more efficiently and apply what it has learned in a changing environment.
Source of the story:
Materials provided by University of Cologne. Note: The content can be edited by style and length.
