A Data-Driven Learning System Based on Natural Intelligence for an IoT Virtual Assistant

A Generic Preprocessing Architecture for Multi-Modal IoT Sensor Data in Artificial General Intelligence

“…Utility functions are currently working at a lower level of abstraction and have corrigibility problems [22,23]. The principle of associative bias aims to solve this problem through a system that is innately impacted by a small set of rudimentary input data characteristics-which grows into a more abstract impact through scene associations and event-chain understanding [24]. This butterfly-like effect enables data impact that begins at a low level at the start of run time, and progresses to higher levels of abstraction, as the system experiences more environments and absorbs more data [24].…”

“…Accordingly, the purpose of this paper is to present a method of input preprocessing conducive to AGI-removing the need for internal intervention such as data labeling input from a human. The preprocessing method presented in this article is a key component of the high-level proposed distributed AGI system presented in our previous work [24]. In this system, as seen in Figure 1, the AGI architecture is spread among wireless, edge, and cloud network components.…”

“…The design of the preprocessing model that aims to target the shortcomings of today's AGI by providing a generic foundation is detailed in the following section. The system model in Section 3 is an extension of a preliminary model presented in our previous work [24], which, however, did not include any proposed algorithms or concrete implementation and evaluations. Hence, in this paper, the results applied to multiple sensor types are also demonstrated and evaluated through a trial run.…”

“…The input alignment and data buffering method proposed in [24] is responsible for structuring sensor data in preparation for system input. The multi-modal data inputted into the system have different frame rates and sample structuring; the input layer structures data into groups according to sample rate and may alter the structure of the data depending on its format.…”

Cloud Computing and Big Data: An Overview

“…The limit and convenience of the current learning systems are reliant upon arrangement and planned for explicit use. These structures serious areas of strength for have on getting ready data and remain static in capacity [8]. By conveying the word through web max of the information are moves carefully by utilizing the various ways [9][10].…”