Hierarchical Temporal DNN and Associative Knowledge Representation

EasyChair Preprint 7921, version 3

Abstract

This paper proposes two models. The first one is designed bottom-up, i.e., mostly based on DL and Jeff Hawkins' temporal principle. The second one tackles some aspects of intelligence, specifically concerning the thinking process. It is designed top-down, i.e., mainly based on cognition and communication.

Additionally, this paper not only exhibits top-down verse bottom-up approaches, but also presents the two edges of evolution: the DL model considers the beginning state of learning, while the knowledge representation model considers the saturated/mature/final state of learning.

Keyphrases: Neuro Science, associative thinking, deep learning

@booklet{EasyChair:7921,
  author    = {Shimon Komarovsky},
  title     = {Hierarchical Temporal DNN and Associative Knowledge Representation},
  howpublished = {EasyChair Preprint 7921},
  year      = {EasyChair, 2022}}