fix template/shortcode here later

Goldhat Hive – Project Hive AI

Overall Project Aims

Create Hive AI – Multiple Node Robots with central power processing unit which provides learning and processing to group.

Schemas

Image result for neural network

Worker- Nodes

  • Single Node Robots – Each Robot Group is independently sufficient
  • Each Nodes has multiple sensors and are several specialised functions.
  • Each Node has on-board storage which backs up to Network Drive

Central Queen 

  • Analyses inputs
  • Co-ordinates overall goals –
  • Distributes updates within groups (Blockchain)
  • Creates bots to investigate goals and suggest solutions
  • Analyses bots results and implement probabilistic suggestions
  • Measure success and adjust configurations

Equipment 

  • Basic single board Connect Create various sensors.

Future development Plans

  • Flexing nodes processing power by rotational configuration through each node becomes primary control node utilitising its groups primary evolutionary focus then learning and trained models distributed.
  • Utilitise configurable routines, virtualenvwrapper and startup protocols to switch between node configuration and development and upgrade depending upon overall aims.

Software

Uses

Sensors 

  • Camera
  • IR
  • Ultrasonic
  • PIR
  • Microphone

Analysis Layers / Software

  • Vision
  • Robot obstacle avoidance
  • OCR
  • Image Recognition
  • Speech Recognition (including speech to text)
  • Text to Speech
  • Wide Semantic Web Searches

 

Definitions of Intelligence:

Following definitions of intelligence are taken from writing of Shane Legg:

  • Intelligence is the property of some active agent.
  • Agent has to interact with the environment
  • Intelligence is a matter of degree – there some kind of scale
  • Intelligence is related to the agents success in achieving goals
  • the environment is not fully known to the agent and so the agent must be adaptable to many different environments
  • Occam’s Razor

He goes on to define these algorithmic at 18.21

Where Pi Is an agent (conditional probability measure)

U is environment

V is success reinforcement learning framework

E wide range of environments

2-K(u) is Occam’s razor (measures complexity of environment and available functions.

 

List of AI Projects

https://en.wikipedia.org/wiki/List_of_artificial_intelligence_projects

Shane Legg – DeepMind – Best PhD thesis I’ve read and one of the most interesting approaches to defining intelligence.

Shane

Machine super intelligence

Legg – 2008 – doc.rero.ch
This thesis concerns the optimal behaviour of agents in unknown computable environments, also known as universal artificial intelligence. These theoretical agents are able to learn to perform optimally in many types of environments. Although they are able to optimally use …

Universal intelligence: A definition of machine intelligence

LeggM Hutter – Minds and Machines, 2007 – Springer
A fundamental problem in artificial intelligence is that nobody really knows what intelligence is. The problem is especially acute when we need to consider artificial systems which are significantly different to humans. In this paper we approach this problem in the following …

Reinforcement Deep Learning –

Great series of videos

part of the project objective

  • OCR
  • Image Recognition
  • Speech Recognition
  • Text to Speech
  • Wide Semantic

 

Goldhat Shallow Iterative Learning Modules – Shallow Minds

Advanced AI Massively parallel processing

Utilities exisiting 3rd party ai

Share findings

 

 

Robotic ai

active ai

reinforcement learning – simulations

 

 

 

Close Menu