Particle swarm optimization using multiple neighborhood connectivity and winner take all activation applied to biophysical models of inferior colliculus neurons
Age-related hearing loss is a prevalent neurological disorder, affecting as many as 63% of adults over the age of 70. The inability to hear and understand speech is a cause of much distress in aged individuals and is becoming a major public health concern as age-related hearing loss has also been correlated with other neurological disorders such as Alzheimer's dementia. The Inferior Colliculus (IC) is a major integrative auditory center, receiving excitatory and inhibitory inputs from several brainstem nuclei. This complex balance of excitation and inhibition gives rise to complex neural responses, which are measured in terms of firing rate as a given parameter is varied. A major obstacle in understanding the mechanisms involved in generating normal and aberrant auditory responses is estimating the strength and tuning of excitatory and inhibitory inputs that are integrated to form the output firing of IC neurons. To better understand IC response generation, biophysically accurate, conductance-based computational models were used to recreate IC frequency tuning responses. The problem of fitting response curves in vivo was approached using particle swarm optimization, an optimization paradigm which mimics social networks of flocking birds to solve problems. A new social network modeling winner-take-all activation found in visual neuron coding was developed in which agents are divided into social hierarchies and compete for leadership rights. This social network has shown good performance in benchmark optimization problems and is used to recreate IC frequency tuning responses which can be used to further understand pathological aging in the auditory system.
Talavage, Purdue University.
Neurosciences|Biomedical engineering|Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our