Towards Generalized Artificial Intelligence by Assessment Aggregation with Applications to Standard and Extreme Classifications

Abdourrahmane M ATTO

Keywords: Artificial Intelligence, Convolutional Neural Network, Extreme Classifications

Abstract: The paper studies the impacts of an aggregative Convolutional Neural Network (CNN) for the derivation of generalized artificial intelligence when starting from a specialized set of CNNs. The aggregation proposed relies on machine learning from specific examples being the categorical probabilities fed by the downstream specialized CNNs under consideration. Aggregation proof of concepts are provided in terms of multimodel, multimodal and distributed schemes. The multimodel framework is such that different CNN models operating on the same modality cooperate for decision purpose. The multimodal framework implies specializations of CNNs with respect to different input modalities. The distributed framework proposed is associated with assessment exchanges: it such that the aggregation aims at determining relevant joint assessments for mapping a given input to a single or a multiple output category. Performance of these aggregation frameworks are shown to be outstanding for both standard and extreme classification issues.

One-sentence Summary: The paper provides an assessment based aggregation framework for the fusion of specialized convolutional neural networks, the motivation being the derivation of an artificial intelligence having more generalization properties.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=El-1sKZmX_