So far nobody has linked together the available building blocks for artificial general intelligence (AGI). These tools include Differentiable Neural Computers, Neural Programmer-Interpreters, Generative Adversarial Networks, and Deep Reinforcement Learning. Together, a single entity that is capable of ingesting multiple types of data, audio and visual recognition, general problem solving, learning to learn, algorithm generation, creativity, and interdisciplinary knowledge integration, may be created, resulting in something sort of like in the movie Her.
Sentience is deployed using Docker as a microservices for each distinct unit of the AGI:
- Executor
Executive functions and cognitive control. Appropriate behaviour selection and operation for goal achievement.
- Interactor
Active communication and interaction with others.
- Strategizer
Data aggregration, analysis, pattern recognition, forecasting, feeding to goal planning, definition of options, preferences, strategies, goals, contexts, and auxillary knowledge store.
An advantage of using Docker is that ensemble learning may be easily used between containers. Apache Kafka is used for messaging between microservices and is also run inside Docker.
a) Docker
b) Compose
b) Git
- Clone the repository:
$ git clone https://github.com/yazanobeidi/sentience.git && cd sentience
- Start with
build.sh(wraps docker-compose) :
$ ./build.sh --prod
- Say hello via the terminal!
- Remove Docker images:
$ docker-compose down --rmi all
- Delete repository:
$ cd .. && rm -r sentience
This project does not attempt to make the claim that strong AI is actually possible or not. What this project does is try to emulate intelligent behaviour using present state-of-the-art tools in the field.
python3.6 -m pip must be used to access pip for Python3.6, as python3.6-pip is not presently packaged with python3.6 in Ubuntu:trusty.
Yazan Obeidi
Yazan Obeidi, 2017-2018
Apache V2
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent