In 2016, the Ministry of Internal Affairs and Communications announced in the 2015 national tax survey that Japan's population is decreasing . In 2018, the Cabinet Office announced Society 5.0, which showed the future of Japan, "A human-centered society that realize both economic development and social issues resolve with the highly synchronized system between cyber space (virtual space) and physical space (real space). On the other hand, in 2015, the United Nations declared the Sustainable Development Goals (SDGs) "17 goals and 169 targets to realize a sustainable world, then leave no one behind". In Japan, in order to realize this future, measures are being taken to improve productivity in companies through work style reforms and to develop the "problem-seeking ability" of university students through educational reforms. This paper proposes next-generation artificial intelligence as the technology that solve Japan domestic problems (securing production methods to compensate for population decline) and exercise Japan leadership to contribute to the achievement of SDGs goals. This is also a proposal for a social model in which new technology realizes the happiness of all people.
(A)Characteristics of memory control of artificial brains with organic information processing that mimics the behavior of the human brain and (B) their application examples are introduced. (1) Structure of associative / hierarchical memory (link connection of memory element) (2) Activation propagation with activation rule of the storage element, (3) Characteristicand property information of the subject / Formation of upper-level concepts, (4) WR: Related information and set in the memory opportunity (activation) when the awareness of the manifestation is off, hierarchical memoryUpdate scene, (5) RD: Storage element activated by external / internal information &Access entities, (6) Hierarchical abstraction control (abstractity control: upper - lower Concept)
In recent years, humanoid robots have become more sophisticated and less expensive, and humanoid robots that download and dance human choreographed dances for a pre-determined song at a very low price It was sold. However, such a robot has a problem that it gets tired after having finished watching all the provided music. Therefore, we have devised a method that can automatically generate a dance motion using a deep generation model for new music data as well as a predetermined piece of music.
We aim at building a model of the information processing mechanism of the prefrontal cortex in the brain. For that purpose, we proposed an architecture of hierarchical reinforcement learning with unlimited recursive subroutine calls, RGoal. In this paper, we show that the slightly extended RGoal can execute a kind of symbolic inference, theorem proving. Moreover, we consider a mechanism of acquiring symbolic knowledge from agent's experience in the environment. These mechanisms are candidates of the models of symbolic inference and knowledge acquisition of human brain.
In the shipbuilding industry, labor shortages are a problem. In particular, designers are required to make flexible decisions with vague information. Therefore, research results of Artificial General Intelligence (AGI) that can collaborate with workers are required. Current machine learning cannot be learned spontaneously. Artificial intelligence never develops while learning behavioral patterns through experience. In this paper, we examined the idea of Deleuze et al to apply self learning like a human behavior. The shipbuilding industry has many challenges to apply AGI.
In this paper, we present a way to distinguish human voice from animal voice. In deep learning-based solutions, we adopted one class classification called DOC and then two class classification called VoVNET to verify all kinds of animal voice and human voice collected on the Internet. We construct a device to react only human voice, starting the voice recognition engine. It has a superior potential when used in a interactive AI in terms of economy. In the result of our study, the accuracy reached about 95% in DOC and over 98% in VoVNET.
The DARPA Robotics Challenge final (DRC final 2015) was the watershed for the trend of world robotics research. The lessons we learned in the DRC final and the following research activities in AIST will be presented. The speaker will emphasize the difficulty of manipulation which is not acknowledged enough by the researchers of Artificial Intelligence.
Bridging between brain structure and its function is crucial issue in the field of neuro science. Many researches have provided the huge amount of knowledges of neuronal connectivities and behaviors. The accumulated knowledges should be summarized in a concept of how neurons behave and orchestrate eaach other to subserve complexed cognitive functions. In this paper, we focused on the neural circuits for saccadic eye movement and facial recognition, and we proposed the ontological approach to accumulate the knowledges of the neural connectivity and its behavior; how cell populations in the neural circuit are interconnected and how they behave from the perspective of role concepts.