There is nothing new in realizing that humans interact with the environment using intelligence, senses and movement/action.
The greatest novelty of the last decades is the desire of these same beings to create machines that emulate themselves or, even, artificially replicate this interaction.
Obviously, we are talking about Artificial Intelligence (AI) and Robotics.
Thus, there are several fields within these areas that in one form or another may exist independently but have grown greatly in recent years with the advancement of Machine Learning technology.
Two well-known examples are Natural Language Processing (NLP) and Computer Vision (CV), both for AI and Robotics, but with independent research and applications.
However, in theory, I believe that all natural data processing, and its representation, could be unified into a single model, which considers any and all processing based on models developed from the observation of nature.
In fact, of course, it is even curious that things are not like that.
And perhaps one of the great problems and delays for the evolution of the Artificial General Intelligence (AGI) area is precisely the lack of a unified view of natural data processing, which I consider to be fundamental for learning, cognitive processing, and general machine vision.
Thus, I propose a new model as being Natural Data Processing (NDP), where any and all sensory model of living beings come to be unified.
For example, sensations perceived by the five senses of the human being, such as sight (vision), hearing (audition), taste (gustation), smell (olfaction), and touch (somatosensation), when modeled on machines, are nothing more than a field based on the single principal model, which is the NDP.
Advantages of the Natural Data Processing (NDP) model
The first direct advantage of this unification is that in practice many of the technologies based on natural data present independent research, often following competing concepts, which does not seem logical to me if all somehow unify in a single larger area.
In addition, we can separate more clearly the areas of Artificial Intelligence and Robotics from the NDP area, since they can be concurrent or complementary, with independent evolution, and whose intersection would seek to emulate in some way all the sciences that contribute to emulate a human being in a machine, that is, an Artificial Being (AB), with the individuality and autonomy of a robot:
Note that in this model, there is no restriction that the robot is anthropomorphic, since the artificial being may not have a specific form or be the result of a collective intelligence.
The most relevant, within the proposed, is to focus the efforts of scientists in the search for AGI in three unified fields, which are AI, NDP and Robotics, representing more directly the human systemic paradigm of interaction with the environment, based on intelligence, sensory and movement/action.
Breaking paradigms through Natural Data Processing (NDP)
One of the most challenging areas for the Computer Vision (CV) field is object recognition/detection.
However, as I propose through a new model called Natural Data Processing (NDP), sensations perceived by the five senses of the human being when modeled on machines, are nothing more than a field based on the main single model, which is the NDP, so, in other words, in the same way that CV can be used for object recognition/detection, analyzing images, we can detect objects through other senses, such as audition and touch, for instance, using NDP.
In the same sense, fields like Natural Language Understanding (NLU) and Natural Language Generation (NLG), within Natural Language Processing (NLP), can be considered only as simple processing of natural data, and their representation, and could be unified in a unique model, which considers any and all processing based on models developed from the observation of nature.
Rogerio Figurelli – @ 2018-12-01