Artificial Intelligence Architecture Inspired by Personality Theory

摘要

This paper introduces a novel approach for classification problems utilizing a Jungian Psychology-inspired classification architecture (JPICA). The goal of JPICA is to demonstrate applications of personality theory to artificial intelligence in general, and thus move closer to holistic artificial intelligence. Here, results are presented for the initial experiments of JPICA applied to the classic AI problem of automated voiced/unvoiced/silence classification of speech segments. Multiple audio files are analyzed, hand labeled, and put through a feature extraction protocol in order to generate the dataset that is used in the experiments. The number of features used in the experiments ranges from 3 to 34. The architecture is tasked with accurately classifying the speech segments, while adhering to the defined behavior of the personalities used to define the various classifiers. Experiments with this early implementation of JPICA provide favorable and encouraging results for future algorithm development and experimentation in the field of artificial intelligence.