Image recognition, Algorithms, Neural networks (Computer science), Artificial intelligence, Nature-inspired algorithms
Two decades since the first convolutional neural network was introduced the AI sub-domains of classification, regression and prediction still rely heavily on a few ML architectures despite their flaws of being hungry for data, time, and high-end hardware while still lacking generality. In order to achieve more general intelligence that can perform one-shot learning, create internal representations, and recognize subtle patterns it is necessary to look for new ML system frameworks. Research on the interface between neuroscience and computational statistics/machine learning has suggested that combined algorithms may increase AI robustness in the same way that separate brain regions specialize. In this paper, a combination of two existing algorithms - a standard multilayer CNN with image parser and Numenta’s spatial pooler unsupervised learning algorithm- is presented to create a system with strong object recognition which also creates internal representations that can be used for predictions and autonomous high level pattern recognition. The system was shown to create detailed and consistent internal representations for images with similar structure but did not recognize high level patterns. Future work in this area should include a front end CNN that captures and encodes highly detailed input information.
Musil, Mark Robert, "Combining Algorithms for More General AI" (2018). Undergraduate Research & Mentoring Program. 20.