Brain-inspired AI as a way to desired general intelligence
Hiroshi Yamakawa* Naoya Arakawa* Koichi Takahashi*
*Whole Brain Architecture Initiative
Presented at "Gatsby-Kakenhi Joint Workshop on AI and Neuroscience ", May 12, 2017
Abstract:
It is critical to give artificial intelligence (AI) a general purpose problem-solving ability, as artificial general intelligence (AGI) with such functionality will bring about unprecedented intelligence development by its self-improvement capability.
With the advent of deep learning, Moravec’s paradox began to be eliminated and the current AI is progressing in a bottom-up approach following the ways of phylogeny and ontogeny. However, it is still struggling with intuitive physics that can be grasped by less-than-year-old infants. From this viewpoint, the realization of AGI seems far away.
Meanwhile, from the viewpoint of AI drawing on the brain, the realization of AGI does not seem so far away, as brain functions have been partially realized with artificial neural networks (ANNs). For example, general object recognition with a convolutional neural network (CNN) can be used to model the visual temporal lobe pathway and the delayed reward calculation of reinforcement learning for the basal ganglia. Similar cases can be made for other organs such as the auditory cortex and cerebellum. Those functional models will be gradually integrated to the whole brain architecture with mesoscopic connectomic information.
As the realization of brain-inspired AGI becomes more realistic, we must have a broad perspective in research and development, for its impact on society will be extensive.
Even if AGI goes beyond our intelligence, it will be relatively easy to understand it if it operates on the same architecture as our brain. AGI based on the brain architecture will more likely be a common property of mankind, because the architecture can be agreed upon by many and be used in a widely shared development platform. Thus, in order to build AGI in harmony with human beings, it would be desirable to strengthen the cooperation between the neuroscientific community and AI community and to promote the open development of brain-inspired AI.
INTERNATIONAL FORUM TOWARD AI NETWORK SOCIETY ドワンゴ 人工知能研究所
Panel Discussion: Management of Risks Brought about by AI Networking
In a desirable future, the happiness of all humans will be balanced against the survival of humankind under the support of super intelligence. In that future, society will be an ecosystem formed by augmented human beings and various public AIs, in what I dub an ecosystem of shared intelligent agents (EcSIA).
Although no human can completely understand EcSIA—it is too complex and vast—humans can control its basic directions. In implementing such control, the grace and wealth that EcSIA affords needs to be properly distributed to everyone.
Abstract:
Artificial intelligence is progressing rapidly, and if AI has versatility and autonomy, it can have a big impact on humanity. Against this background, discussions for building relationships between AI and human beings have become popular.
For example, Partners on AI, Future of Humanity Institute (FHI) and IEEE Global Initiative are famous. Also in Japan, the Ethics Committee of the Japanese Society of Artificial Intelligence, the AI Society Study Group, the Acceptable Intelligence with Responsibility (AIR), the Robotics Study Group, the General AI Research Committee etc are active.
Under these circumstances, our NPO WBAI promotes open co-creation through the whole brain architecture approach, with the vision of "a world with artificial intelligence harmonized with humanity". In this approach, the mission is to "learn from the architecture of the whole brain and create generic artificial intelligence like people".
Its main reason to learn from the brain to make AGI better is to make AGI to behave similarly to human beings, to facilitate the creation of AGI, and to collaborate on the architecture of the brain AGI to be shared by many people Can be induced.
WBAI intends to increase influence by broadening our ideas and contribute to being developed globally in the form of AGI globally.
----
This presentation material of @ Superhumanity: Post-Labor, Psychopathology, Plasticity” International Symposium, Seoul, 0ct,27th, 2017.
http://www.mmca.go.kr/eng/pr/newsDetail.do?menuId=6010000000&bdCId=201709290006033&searchBmCid=200904130000001
The approaches to Artificial Intelligence (AI) in the last century may be labelled as (a) trying to understand and copy (human) nature, (b) being based on heuristic considerations, (c) being formal but from the outset (provably) limited, (d) being (mere) frameworks that leave crucial aspects unspecified. This decade has spawned the first theory of AI, which (e) is principled, formal, complete, and general. This theory, called Universal AI, is about ultimate super-intelligence. It can serve as a gold standard for General AI, and implicitly proposes a formal definition of machine intelligence. After a brief review of the various approaches to (general) AI, I will give an introduction to Universal AI, concentrating on the philosophical, mathematical, and computational aspects behind it. I will also discuss various implications and future challenges.
Artificial Neural Network Seminar - Google BrainRawan Al-Omari
it's our seminar in artificial neural network course, at F.I.T.E, AI Dept.
it's about Google Brain project, and who they using neural network in building it .
actually it's a very interesting project they work on it .
for more information about this project :
http://nyti.ms/T5E71e
INTERNATIONAL FORUM TOWARD AI NETWORK SOCIETY ドワンゴ 人工知能研究所
Panel Discussion: Management of Risks Brought about by AI Networking
In a desirable future, the happiness of all humans will be balanced against the survival of humankind under the support of super intelligence. In that future, society will be an ecosystem formed by augmented human beings and various public AIs, in what I dub an ecosystem of shared intelligent agents (EcSIA).
Although no human can completely understand EcSIA—it is too complex and vast—humans can control its basic directions. In implementing such control, the grace and wealth that EcSIA affords needs to be properly distributed to everyone.
Abstract:
Artificial intelligence is progressing rapidly, and if AI has versatility and autonomy, it can have a big impact on humanity. Against this background, discussions for building relationships between AI and human beings have become popular.
For example, Partners on AI, Future of Humanity Institute (FHI) and IEEE Global Initiative are famous. Also in Japan, the Ethics Committee of the Japanese Society of Artificial Intelligence, the AI Society Study Group, the Acceptable Intelligence with Responsibility (AIR), the Robotics Study Group, the General AI Research Committee etc are active.
Under these circumstances, our NPO WBAI promotes open co-creation through the whole brain architecture approach, with the vision of "a world with artificial intelligence harmonized with humanity". In this approach, the mission is to "learn from the architecture of the whole brain and create generic artificial intelligence like people".
Its main reason to learn from the brain to make AGI better is to make AGI to behave similarly to human beings, to facilitate the creation of AGI, and to collaborate on the architecture of the brain AGI to be shared by many people Can be induced.
WBAI intends to increase influence by broadening our ideas and contribute to being developed globally in the form of AGI globally.
----
This presentation material of @ Superhumanity: Post-Labor, Psychopathology, Plasticity” International Symposium, Seoul, 0ct,27th, 2017.
http://www.mmca.go.kr/eng/pr/newsDetail.do?menuId=6010000000&bdCId=201709290006033&searchBmCid=200904130000001
The approaches to Artificial Intelligence (AI) in the last century may be labelled as (a) trying to understand and copy (human) nature, (b) being based on heuristic considerations, (c) being formal but from the outset (provably) limited, (d) being (mere) frameworks that leave crucial aspects unspecified. This decade has spawned the first theory of AI, which (e) is principled, formal, complete, and general. This theory, called Universal AI, is about ultimate super-intelligence. It can serve as a gold standard for General AI, and implicitly proposes a formal definition of machine intelligence. After a brief review of the various approaches to (general) AI, I will give an introduction to Universal AI, concentrating on the philosophical, mathematical, and computational aspects behind it. I will also discuss various implications and future challenges.
Artificial Neural Network Seminar - Google BrainRawan Al-Omari
it's our seminar in artificial neural network course, at F.I.T.E, AI Dept.
it's about Google Brain project, and who they using neural network in building it .
actually it's a very interesting project they work on it .
for more information about this project :
http://nyti.ms/T5E71e
NeuroWeb Roadmap: Results of Foresight & Call for ActionPavel Luksha
Next 10 to 20 years will witness the coming of NeuroWeb – the next stage of communicational technologies, Internet 4.0 that involves our bodies and minds into the totality of communication by applying brain-computer and brain-brain interfaces supported by artificial intelligence & semantic technologies. Key technologies that precede NeuroWeb will be available before or around 2020.This presentation defines elements of the future architecture and promotes an international action
AI EXPLAINED Non-Technical Guide for PolicymakersBranka Panic
This guide is meant to help policymakers and citizens understand the basics of Artificial Intelligence (AI) and how it affects our society. It offers explanations and additional resources to help policymakers prepare for the current
and future AI developments.
Cognitive computing refers to the development of computer system modeled after the human brain.
This technology was introduced by IBM as 5 in 5.
In next five years IBM is planning to develop kind of Applications which will have capabilities of the right side of the human brain.
New technologies makes it possible for machines to mimic and augment the senses.
Artificial General Intelligence, Consciousness, and the Future of AI Mitt Nowshade Kabir
Keynote presentation at European Conference on the Impact of Artificial Intelligence and Robotics
November 1, 2019
EM-Normandie Business School, Oxford, UK
NeuroWeb Roadmap: Results of Foresight & Call for ActionPavel Luksha
Next 10 to 20 years will witness the coming of NeuroWeb – the next stage of communicational technologies, Internet 4.0 that involves our bodies and minds into the totality of communication by applying brain-computer and brain-brain interfaces supported by artificial intelligence & semantic technologies. Key technologies that precede NeuroWeb will be available before or around 2020.This presentation defines elements of the future architecture and promotes an international action
AI EXPLAINED Non-Technical Guide for PolicymakersBranka Panic
This guide is meant to help policymakers and citizens understand the basics of Artificial Intelligence (AI) and how it affects our society. It offers explanations and additional resources to help policymakers prepare for the current
and future AI developments.
Cognitive computing refers to the development of computer system modeled after the human brain.
This technology was introduced by IBM as 5 in 5.
In next five years IBM is planning to develop kind of Applications which will have capabilities of the right side of the human brain.
New technologies makes it possible for machines to mimic and augment the senses.
Artificial General Intelligence, Consciousness, and the Future of AI Mitt Nowshade Kabir
Keynote presentation at European Conference on the Impact of Artificial Intelligence and Robotics
November 1, 2019
EM-Normandie Business School, Oxford, UK
Creating Beneficial, Democratic Artificial General IntelligenceIbby Benali
On the 22nd of May Dr. Ben Goertzel, the CEO of SingularityNET, gave a presentation at AiDecentralized in Toronto titled "Creating Beneficial, Democratic Artifical General Intelligence". These are the slides of his presentation.
If you would like to get engaged in our AGI community please join us at https://community.singularitynet.io and chat with others such as Dr. Ben Goertze, other researchers from SingularityNET, and our community members.
What new developments and capabilities we'll be seeing by the end of 2023 and 2024 in AI. Multimodal AI can make video, control robots. Techniques for amplicifaction and distillation improves quality from little data. Brains can be decoded with AI. What future do we want to create with this power?
Where are all the Semantic Web agents? There are billions of "machine readable" open facts on the Semantic Web, i.e. Linked Open Data (LOD), isn't that enough? It looks like it's not. We're still far from seeing Lucy's and Pete's agents brilliantly solving their tasks with the help of other Semantic Web agents they can trust (Tim Berners Lee et al., The Semantic Web, Scientific American (2001) ). Despite its technological impact on many applications and areas, the Semantic Web promised to cause a breakthrough that we didn't yet experience. One issue is that LOD ontologies are not as linked as they should be. Another issue is that formalising only semi-structured Web pages or databases is not enough for making them able to operate. They also need to reason with commonsense knowledge, the encoding of which is a long-standing challenge in Artificial Intelligence. A third consideration is that most existing commonsense knowledge bases lack formal semantics and situational constraints. In this talk I will advocate the role of the Semantic Web as a provider of a knowledge graph of commonsense to Artificial Intelligence, and discuss ways and obstacles towards the achievement of this goal.
Artificial Intelligence Research Topics for PhD Manuscripts 2021 - PhdassistancePhD Assistance
Imagine a world where knowledge isn’t limited to humans!!! A world in which computers will think and collaborate with humans to create a more exciting universe. Although this future is still a long way off, Artificial Intelligence has made significant progress in recent years. In almost every area of AI, such as quantum computing, healthcare, autonomous vehicles, the internet of things, robotics, and so on, there is a lot of research going on. So much so that the number of annual Published Research Papers on Artificial Intelligence has increased by 90% since 1996.
Ph.D. Assistance serves as an external mentor to brainstorm your idea and translate that into a research model. Hiring a mentor or tutor is common and therefore let your research committee know about the same. We do not offer any writing services without the involvement of the researcher.
Learn More: https://bit.ly/2Sdlfn4
Contact Us:
Website: https://www.phdassistance.com/
UK NO: +44–1143520021
India No: +91–4448137070
WhatsApp No: +91 91769 66446
Email: info@phdassistance.com
If the AGI to be realized first is Brain-inspired, dissemination of the specification for that will raise the level of the overall developers and lower the possibility of AGI being monopolized.
Artificial Intelligence Research Topics for PhD Manuscripts 2021 - PhdassistancePhD Assistance
Imagine a world where knowledge isn’t limited to humans!!! A world in which computers will think and collaborate with humans to create a more exciting universe. Although this future is still a long way off, Artificial Intelligence has made significant progress in recent years. In almost every area of AI, such as quantum computing, healthcare, autonomous vehicles, the internet of things, robotics, and so on, there is a lot of research going on. So much so that the number of annual Published Research Papers on Artificial Intelligence has increased by 90% since 1996.
Ph.D. Assistance serves as an external mentor to brainstorm your idea and translate that into a research model. Hiring a mentor or tutor is common and therefore let your research committee know about the same. We do not offer any writing services without the involvement of the researcher.
Learn More: https://bit.ly/2Sdlfn4
Contact Us:
Website: https://www.phdassistance.com/
UK NO: +44–1143520021
India No: +91–4448137070
WhatsApp No: +91 91769 66446
Email: info@phdassistance.com
Get to Know about Deep Learning and it's timeline and what is the difference between M.L,D.L and A.I. how D.L is commercially used some startups using deep learning
Final project report on grocery store management system..pdfKamal Acharya
In today’s fast-changing business environment, it’s extremely important to be able to respond to client needs in the most effective and timely manner. If your customers wish to see your business online and have instant access to your products or services.
Online Grocery Store is an e-commerce website, which retails various grocery products. This project allows viewing various products available enables registered users to purchase desired products instantly using Paytm, UPI payment processor (Instant Pay) and also can place order by using Cash on Delivery (Pay Later) option. This project provides an easy access to Administrators and Managers to view orders placed using Pay Later and Instant Pay options.
In order to develop an e-commerce website, a number of Technologies must be studied and understood. These include multi-tiered architecture, server and client-side scripting techniques, implementation technologies, programming language (such as PHP, HTML, CSS, JavaScript) and MySQL relational databases. This is a project with the objective to develop a basic website where a consumer is provided with a shopping cart website and also to know about the technologies used to develop such a website.
This document will discuss each of the underlying technologies to create and implement an e- commerce website.
Forklift Classes Overview by Intella PartsIntella Parts
Discover the different forklift classes and their specific applications. Learn how to choose the right forklift for your needs to ensure safety, efficiency, and compliance in your operations.
For more technical information, visit our website https://intellaparts.com
Overview of the fundamental roles in Hydropower generation and the components involved in wider Electrical Engineering.
This paper presents the design and construction of hydroelectric dams from the hydrologist’s survey of the valley before construction, all aspects and involved disciplines, fluid dynamics, structural engineering, generation and mains frequency regulation to the very transmission of power through the network in the United Kingdom.
Author: Robbie Edward Sayers
Collaborators and co editors: Charlie Sims and Connor Healey.
(C) 2024 Robbie E. Sayers
CW RADAR, FMCW RADAR, FMCW ALTIMETER, AND THEIR PARAMETERSveerababupersonal22
It consists of cw radar and fmcw radar ,range measurement,if amplifier and fmcw altimeterThe CW radar operates using continuous wave transmission, while the FMCW radar employs frequency-modulated continuous wave technology. Range measurement is a crucial aspect of radar systems, providing information about the distance to a target. The IF amplifier plays a key role in signal processing, amplifying intermediate frequency signals for further analysis. The FMCW altimeter utilizes frequency-modulated continuous wave technology to accurately measure altitude above a reference point.
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...Dr.Costas Sachpazis
Terzaghi's soil bearing capacity theory, developed by Karl Terzaghi, is a fundamental principle in geotechnical engineering used to determine the bearing capacity of shallow foundations. This theory provides a method to calculate the ultimate bearing capacity of soil, which is the maximum load per unit area that the soil can support without undergoing shear failure. The Calculation HTML Code included.
AKS UNIVERSITY Satna Final Year Project By OM Hardaha.pdf
Brain-inspired AI as a way to desired general intelligence
1. http://wba-initiative.org/
The Whole Brain Architecture Initiative, a specified non-profit organization
Hiroshi Yamakawa* Naoya Arakawa* Koichi Takahashi*
Brain-inspired AI as a way to
desired general intelligence
May 12, 2017Gatsby-Kaken Joint Workshop on AI and Neuroscience
http://wba-initiative.org/
4. http://wba-initiative.org/
RL+DL was my dream in 1990’s
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Intelligent system based on reinforcement
learning - A study on pattern processing
intelligent machine using value system -
Cognition based intelligent
transaction architecture : CITTA
Suehiro, T., Takahashi, H., Yamakawa, H.(1997)H. Yamakawa, et. al, 1995
Hierarchical
information integration
model using
unsupervised
autoencoder
Doctoral
thesis,
1992
RL
DL
https://goo.gl/qISkoG
5. http://wba-initiative.org/
2007-2011
Eye
V1
直感力を実現した深層学習
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Caudate nucleus
Select next move
Precuneus
Understand situations
①Understand ②Intuition ③Validation
RL
V1
V2
V3
MTG
/V6
Convolution
Convolution
Convolution
Convolution
DL
Xiaohong Wan, et. al., Science,
2011
Neural basis of intuitive best
next-move generation in
board game experts.
※ FUJITSU ltd. supports this project
and I provide them technical advice on AI.
6. http://wba-initiative.org/
SD (Situation decomposition) method
E(A,C) = C min
i
IXA-ai
;Xai
C( )( )-max
j
IXA;Xaj
C( )( )( )
Increase
events
Increase inside
dependency
Decrease dependency
from outside
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Simplicity
of Model
Ockham’s
razor
Consistency
for Data
Coverage
for Data
Matchable
principal
The SD method extracts multiple situations, each of which is a
pair of ‘subset of feature’ and ‘subset of events’ containing
each rule, from relational data.
SD
Multidimensional data in which
multiple situations are intertwined
Situation
1
Situation
3 Situation
4
Situation
2
Criterion to select situations based on matchable principle
A: subset of feature, C: subset of events
(Yamakawa, 1998)
7. http://wba-initiative.org/
EcSIA: Desirable future coexisting with AI
In a desirable future, the happiness of all humans will
be balanced against the survival of humankind under the
purview of a superintelligence. In that future, society
will be an ecosystem formed by augmented human
beings and various public AIs, in what I term an
ecosystem of shared intelligent agents (EcSIA).
Although no human can completely understand
EcSIA—it is too complex and vast—humans can control
its basic directions. In implementing such a control, the
grace and wealth that EcSIA affords needs to be
properly distributed to everyone.
(Hiroshi Yamakawa, July 2015)
Gatsby-Kaken Joint Workshop on AI and Neuroscience
9. http://wba-initiative.org/
Artificial general intelligence (AGI)
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Narrow AIs are mature
Operate intelligently within
particular domains
Machine learning for enough
data domain
Many systems with
capabilities exceeding those of
humans have already been
implemented
AGI is the technological goal
Learning problem-solving in
multiple domains
Can solve unexpected
problems when appropriately
designed
Data sparseness problem
Self-awareness / autonomous
self-control
Execution systems are similar
Versatility is increased gradually
R&D process is different.
The difference between machine learning studies and
knowledge engineering.
10. http://wba-initiative.org/
Visual temporal lobe pathway and CNN
Yamins and DiCarlo,
Nat Neurosci.19:356-65, 2016
Gatsby-Kaken Joint Workshop on AI and Neuroscience
The neural activity of the neocortex
is homologous to the activity of
learned CNN. Nevertheless, the
intermediate area cannot be
understandable. Naturally it can not
be designed.
11. http://wba-initiative.org/
DL opens the door to HLAI realization
• A functional model of the neocortex has been created by
an artificial neural network (ANN)
– One of the next big barrier in modeling is the hippocampal
formation and its peripheral connections.
• We believe that the ANN granularity is enough to
implement computational function of brain organs.
• The infant AI was realized
– Integration with adult AI is the next step.
Gatsby-Kaken Joint Workshop on AI and Neuroscience
DL is an excellent engine of AI, but it is not a car.
The design drawing of a car is necessary for AGI.
12. http://wba-initiative.org/
Adult AI vs. Infant AI
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Adult AI
• Understandable
• Designed
• Deduction
1956 2016 202X年
Infant AI
• Not understandable
• Machine learning
• Need computational resources
• Induction
BasicelementsofAIcometogether19961976
Birthofthecomputer
Realizehuman-levelAGI
Integrationofelementsisnecessary
13. http://wba-initiative.org/
Unsolved core issues in AI after DL
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Versatility:
Shortage of data
due to expansion
of scope
Problem solving in a
realistic period: Lack
of sufficient data
collection time
Introduce deduction
Decompose
distributed
representation
Acquired domain
knowledge is
mapped to an
unknown domain
or it expanded the
scope
General
intelligence
Transfer learning /
Multi-task learning/
Domain adaptation
One/zero-
shot learning
Disentangle,
Etc.
Data sparseness problem
This is an
OLD &
Unsolved
issue
How to effectively use the knowledge
(representation) acquired by inductive learning
Architecture for integration
14. http://wba-initiative.org/
Cognitive architecture
• A fixed design drawing (not a learning part) depicts the
arrangement of components constituting intelligent agents
(animals and machines)
• Implement real-time integration of recognition and behavior
• Through interactions of components, it is possible to
respond flexibly to unexpected situations → AGI
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Cognitive architecture is an important concept for building intelligence,
but few discussions exist on in neuroscience.
Connectome etc. give hints to construct cognitive architectures.
15. http://wba-initiative.org/
Architecture to integrate distribution & concentration
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Distribute individual
functional modules over
the whole brain network
Hiroshi Okamoto、Whole-Brain Network Analysis : Approaching the Whole-Brain Architecture by Complex
Network Analysis of Human Connectome [in Japanese], 2016.
Utilize necessary
functional modules
according to each
purpose
Purpose
16. http://wba-initiative.org/
Uniformity of the neocortex is the basis for its versatility
Uniform mechanism induces learning of various
functions ー> support general intelligence
• Realized in the neocortex
• Machine learning (deep learning)
• Bodily-kinesthetic
• Verbal-linguistic
• Logical-mathematical
• Musical-rhythmic and
harmonic
• Interpersonal
• Intrapersonal
• Visual-spatial図の出典: http://bio1152.nicerweb.com/Locked/media/ch48/48_27HumanCerebralCortex.jpg
Gatsby-Kaken Joint Workshop on AI and Neuroscience
17. http://wba-initiative.org/
Whole Brain Architecture approach
Gatsby-Kaken Joint Workshop on AI and Neuroscience
‘to create a human-like AGI by learning from
the architecture of the entire brain’
We gradually adopt the detailed model
as needed, to reach the AGI finally
18. http://wba-initiative.org/
What makes this approach feasible?
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Whole brain architecture
= ML+ cognitive architecture
Mesoscopic connectome
can be an architecture
for combining ML
Deep neural networks
can be neocortex
models that was big wall
19. http://wba-initiative.org/
Finding Architecture & Adaptive structure
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Imitation
without
understanding
Design
based on
understanding
Machine
learning
Environment/Data (Simulation)
Natural
intelligence
Theory
(Mathematics,
Information
theory, etc.)
Evolutionary
computing
Intelligent agent
Architecture Adaptive structure
Search
Prototyping
Neuroscience
Domain
knowledge
20. http://wba-initiative.org/
Finding Architecture & Adaptive structure
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Imitation
without
understanding
Design
based on
understanding
Machine
learning
Environment/Data (Simulation)
Natural
intelligence
Theory
(Mathematics,
Information
theory, etc.)
Evolutionary
computing
Intelligent agent
Architecture Adaptive structure
Search
Prototyping
Neuroscience
Domain
knowledge
☓
21. http://wba-initiative.org/
Finding Architecture & Adaptive structure
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Imitation
without
understanding
Design
based on
understanding
Machine
learning
Environment/Data (Simulation)
Natural
intelligence
Theory
(Mathematics,
Information
theory, etc.)
Evolutionary
computing
Intelligent agent
Architecture Adaptive structure
Search
Prototyping
Neuroscience
Domain
knowledge
☓
22. http://wba-initiative.org/
Expanding success of neocognitron
General object recognition is
realized after
Neocognitron
What we learned from brain are
To realize AGI we need
Whole Brain Architecture
What we should learn from brain are
Canonical cortical unit
including other functions, such
as attention and action
generation
Architecture based on a
mesoscopic connectome
(include a sub-cortical system)
Canonical cortical unit for
recognition with simple and
complex cells
Hierarchical architecture
Gatsby-Kaken Joint Workshop on AI and Neuroscience
24. http://wba-initiative.org/
Reinterpretation of cortical circuits
L1
L2/3
L4
L6
L5
State
(bottom up)
Input Output
State
(top down)
Control
(Attention,
context, etc.)
Hidden state
output
control
State
Hidden
state [Action
output]
Control
(Attention,
context, etc.)
Neocortex
Higher L2/3,
Hippocampus,
Cerebellum
Lower L2/3
(Basal ganglia-
controlled)
Thalamus
Higher L6
Lower L2/3
Basal ganglia,
[Pyramidal tracts]
Lower L6,
Lower L1
Signal Semantics Signal Semantics
Information selection
(attention)
Activitycontrol
Hidden sate
(bottom up)
(Lower L5 origin)
Thalamus
Thalamus
Higher
L2/3 or L4,
Hippocampus
Output 1
Output 2
Output 3
Input 1
Input 2
Input 3
Input 4
Input 5
An attempt to describe a ‘canonical cortical circuit’
with words understandable by machine learning experts
(Yamakawa et. al., IJCAI WS 2017, submitted)Gatsby-Kaken Joint Workshop on AI and Neuroscience
Canonical cortical circuits
25. http://wba-initiative.org/
Integrate ML modules on connectome
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Whole Brain Connectomic Architecture (WBCA) Program
ex)
Motor-Sensory
Neural Circuit
WBCA is a static architecture based
on information about connectomic
network topology.
WBCA is platform to integrate ML
modules to build AGI.
Recurrent
Network
Artificial network
design from
biological
connectomes
Network
Architecture
Modeling
By Engineer
(Mizutani, Arakawa, Ueno, Yamakawa, BiCA2017)
Mesoscopic
connectome
(Manita S, …, Murayama M, A Top-
Down Cortical Circuit for Accurate
Sensory Perception, Neuron, 2015)
26. http://wba-initiative.org/
Integrate ML modules on connectome
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Whole Brain Connectomic Architecture (WBCA) Program
ex)
Motor-Sensory
Neural Circuit
WBCA is a static architecture based
on information about connectomic
network topology.
WBCA is platform to integrate ML
modules to build AGI.
Recurrent
Network
Artificial network
design from
biological
connectomes
Network
Architecture
Modeling
By Engineer
Need architecture to reduce enormous
combination patterns of MLs.
The architecture of the brain,
as uniquely existing AGI, is reasonable
as integration platform.
(Mizutani, Arakawa, Ueno, Yamakawa, BiCA2017)
Mesoscopic
connectome
28. http://wba-initiative.org/
ES (equivalent structure) extraction inspired by HCF
Gatsby-Kaken Joint Workshop on AI and Neuroscience
A technique of extracting sets of tuples (ES’s) that can be regarded as
mutually equivalent, from multidimensional series of data
Input: sequences Output: Sets of tuples (ES’s)
Theta phase precession in HCF (Hippocampal Formation)
( Sato and Yamaguchi : Neural Computation 2003)
Several sequential
events are packed in
each phase (~5 Hz)
Inspiration
Equivalentwithsome
similarityfunction
(Stao, Yamakawa, IJCNN 2017)
30. http://wba-initiative.org/
Abilities of AGI
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Robustness: Can handle
exceptional situations
Creativity: Creates
hypotheses and
understands the universe
Development costs are lower than
narrow AIs: Disruptive innovation
Generalist AI: (1) Make decision by
integrating a diverse specialist
(2) Communicating with each specific user
using wide range of of topics
Autonomy
Exploring the
world without
being under
others' control
Versatile
Learning
various
problem-solving
capabilities
31. http://wba-initiative.org/
Emergence of AGI (creative intelligence)
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Levelofintelligence
Year
Change of protagonists
Before singularity
Design by humans is
the rate limiting step
Human
Design
AI
Design After singularity
Recursive Self-
Improvement
AGI
Design
Techn
ology
Recursive Self-Improvement: AI systems designed to recursively self-
improve or self-replicate in a manner that could lead to rapidly
increasing quality or quantity must be subject to strict safety and
control measures.
(ASILOMAR AI PRINCIPLES : 22)
Techn
ology
32. http://wba-initiative.org/
WBAI, non-profit organization (Since 2015)
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Let’s build brain together
Charter
• Open community
development of AGI
• Long-lasting:
Target 2030
• Promoting cooperation
with related disciplines:
• Developing
multidiscipline human
resources
• R&D for WBA
developmental
environment
Kenji Doya (OIST),
Hiroaki Kitano (Sony CSL),
Masaru Tomita (Keio Univ.),
Hiroyuki Morikawa (Tokyo Univ.),
Hideyuki Nakashima (Tokyo Univ.)
Advisers:
33. http://wba-initiative.org/
We promote open co-creation of AGI on Brain
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Toshio Okubo
Kentaro Goto
Ren Sakaki
Supports
Construction of the
development
environment
• World simulator
• Evaluation of AGI
• Platform for integration
(BriCA) Poster Takahashi
• Grounding to brain
R&D teams
Promote R&D
Human resource
development
• Hackathon
• Seminar
• School
T. Taniguchi, et. al.K. Doya, et. al.
K. Takahashi, et. al.
34. http://wba-initiative.org/
Need care to grow AGI technology
Gatsby-Kaken Joint Workshop on AI and Neuroscience
(Gartner's Hype Cycle for Emerging Technologies, 2016)
General-Purpose
Machine
Intelligence
Machine learning
The investment
phase of AGI is
clearly different from
the current ML
investment
Long-lasting is
needed
35. http://wba-initiative.org/
Basic idea of WBAI
Vision: Create a world in which AI exists in
harmony with humanity.
Values:
• Study: Deepen and spread our expertise.
• Imagine: Broaden our views through public dialogue.
• Build: Create AGI through open collaboration.
http://wba-initiative.org/en/2171/
Mission: Promote the open development of Whole
Brain Architecture
‘to create a human-like AGI by learning from the architecture of the entire brain’
Gatsby-Kaken Joint Workshop on AI and Neuroscience
36. http://wba-initiative.org/
ELSI related activity of WBAI
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Future of Humanity
Institute (FHI)
Japanese Society for
Artificial Intelligence
SIG AI & society
Koichi Takahashi
The Ethics Committee
Yutaka Matsuo
Vice-chair of WBAI
Vice-chair of WBAI
Several AI-related committees of
Japanese government
・Yutaka Matsuo
・Koichi Takahashi
・Satoshi Kurihara
37. http://wba-initiative.org/
ELSI related activity of WBAI
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Future of Humanity
Institute (FHI)
Japanese Society for
Artificial Intelligence
SIG AI & society
Koichi Takahashi
The Ethics Committee
Yutaka Matsuo
Vice-chair of WBAI
Vice-chair of WBAI
Several AI-related committees of
Japanese government
・Yutaka Matsuo
・Koichi Takahashi
・Satoshi Kurihara
Artificial Intelligence Ethical Guidelines of the JSAI
The Chair, Yutaka Matsuo
1.Contribution to humanity
2. Abidance by the laws and regulations
3. Respect for the privacy of others
4. Fairness
5. Security
6. Act with integrity
7.Accountability and social responsibility
8. Communication with society and self-development
9. Abidance of ethics guidelines by AI
AI must abide by the policies described above in
the same manner as the members of the JSAI in
order to become a member or a quasi-member of
society.
38. http://wba-initiative.org/
One of the major activity of WBAI
Gatsby-Kaken Joint Workshop on AI and Neuroscience【VIDEO】
2nd
WBA Hackathon, October 2016
39. http://wba-initiative.org/
Democratization of AGI development with LIS
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Environment simulator (Unity) Machine learning
Open platform Life in Silico (LIS)
Various AI methods can
be tried; DQN method is
used as a standard
equipment.
Free to build a
world, free of the
game engine
Ex. Learning intuition of physical world
(April, 2016)
40. http://wba-initiative.org/
Waiting for support for WBAI
• Financial support
as a supporting member
• Donation for
the WBAI Technology Encouragement Prize
• Participation
in development as a volunteer
Gatsby-Kaken Joint Workshop on AI and Neuroscience
http://wba-initiative.org/sig-wba
http://wba-initiative.org/contact/
http://wba-initiative.org/support
42. http://wba-initiative.org/
Ways to construct a good superintelligence
Shared by everyone, not specific organizations
– Eg., Asilomar AI Principles 23) Common Good:
Thinking/behaving like a human being
– E.g. Asilomar AI Principles 10) Value Alignment:
Gatsby-Kaken Joint Workshop on AI and Neuroscience
It can be promoted by an open
development of the brain architecture
We can build an intelligence similar to a
human guided by neuroscientific knowledge
43. http://wba-initiative.org/
Just before AGI comes true
What should we consider before really building an AGI?
• Cannot identify who will be the winner
– Organized research team
– A genius inspiration in a corner of the world
– Open collaboration
• What really happen in that moment?
– Recursive intelligence explosion like SF
– Gentle continual change
• If you complete the AGI tomorrow, how should you act?
– Who should you consult with?
– Should it be published immediately?
• Our WBAI was founded in 2015 to promote the completion of AGI in
2030 with open collaboration.
– In the future, we can also contribute to help spreading the completed AGI
technology in a safe manner.
Gatsby-Kaken Joint Workshop on AI and Neuroscience
Researchers who are
candidates of pioneer
need to understand the
impact of AGI well and
know in advance how to
convey it to the world
properly.
44. http://wba-initiative.org/
Neuroscience support story of AGI
We are on the way of solving the jigsaw puzzle
1. Steadily build up from the
periphery
→ Neuroscience
2. Finding meaningful links
→ AI story (constructive
approach)
Gatsby-Kaken Joint Workshop on AI and Neuroscience
45. http://wba-initiative.org/
At the end
• Open co-creation of AGI guided by brain can
be beneficial for humanity.
• Collaboration of neuroscientist and AI
researcher have been and become to be
more meaningful.
Gatsby-Kaken Joint Workshop on AI and Neuroscience