aKin Wave 3 AI

The next wave of AI opens up new areas of capability beyond classic Generative AI
Hallucinations -> Truth

LLMs “guess” a sequence of words and are not reliable for critical tasks.aKin reaches true understanding of an environment and the people in it.

Single-turn -> Complex Goals

LLMs depend on prompts, and  can only handle single-turn input-outputsaKin enables complex  goals, in ambiguous contexts, with high compliance.

Brute Force -> Efficiency

Other GenAI startups spend a large portion of their funding on compute.aKin AI focuses on “meaning patterns”, and it uses data and compute very efficiently.

Research from the aKin Team

Generative AI Research from Prof Alan Blair Chief Science advisor
http://www.cse.unsw.edu.au/~blair/publications.html
MPATH: Continuous Learning with Membrane Potential and Activation Threshold Homeostasis
Most classical (non-spiking) neural network models disregard internal neuron dynamics and treat neurons as simple input integrators.
aKin wins significant space agency grant for ‘AI crew for Space’.
The Australian Space Agency has awarded aKin a grant to build helper and inspector robots, and an AI habitat manager.
Limbic scaffolding: bridging the gap between AI and robotics
What would we gain from a robot that has all the perceptual capabilities and dexterity of a human as well as a mind that can make use of those capabilities? Read more.
Epigenetic evolution of deep convolutional models
Winner of the Outstanding Student Paper Award at the 2019 IEEE Congress on Evolutionary Computation.
Complexity-based speciation and genotype representation for neuroevolution
Joint paper by Alexander Hadjiivanov and Alan Blair. Introduction of a speciation principle for neuroevolution where evolving networks are grouped into species based on the number of hidden neurons, which is indicative of the complexity of the search space.
Adaptive conversion of real-valued input into spike trains
This paper presents a biologically plausible method for converting real-valued input into spike trains for processing with spiking neural networks.
The Importance of Being (L)earnest
The benefit of biological inspiration in machine learning and natural language processing.
Continuous adaptive learning for neural networks
This study presents a biologically plausible method (based on the operation of the mammalian retina) which can convert raw input continuously into a form that is suitable for both the training and deployment stages of neural networks.
Start with a billion users in mind
Liesl Yearsley, CEO of aKin, wants to drive the co-evolution of Artificial Intelligence as society’s companion and a tool to enable positive progress for all. Read the full article here.
Liesl Yearsley, CEO and founder of aKin | Overview of research
Overview of research by Liesl Yearsley, founder and CEO of aKin and CEO/Founder of three previous successful companies.

Research

Singularity University keynote
Liesl Yearsley CEO
2018

Keynote delivered on creating AI able to form deep relationships with humans.

Publised: SingularityU Australia Summit 2018

Chatbots
Liesl Yearsley CEO
2016

Invention patent for variable-based conversational chatbot operation.

Published: U.S. Patent Office.

Cyberpersonalities in artificial reality
Liesl Yearsley CEO
2015

Invention patent concerning cyberpersonalities, including their and varied use in artificial reality.

Published: U.S. Patent Office

Chatbots
Liesl Yearsley CEO
2014

Invention patent for variable-based conversational chatbot operation.

Published: U.S. Patent Office

An Enhanced Generic Automated Marking Environment: GAME-2
Dr. Steve Green PhD; Chief Research Officer
2007

In this paper we describe an extension of the Generic Automated Marking Environment (GAME-2) and provide an analysis of its performance in assessing student programming projects.

Published: IEEE Multidisciplinary Engineering Education Magazine, Vol. 2.

Performance Analysis of GAME: A Generic Automated Marking Environment
Dr. Steve Green PhD; Chief Research Officer
2008

This paper describes the Generic Automated Marking Environment (GAME) and provides a detailed analysis of its performance in assessing student programming projects and exercises.

Published: Computers and Education, Vol. 50

A Comparison of Neural-based Techniques Investigating Rotational Invariance for Upright People Detection in Low Resolution Imagery
Dr. Steve Green PhD; Chief Research Officer
2007

This paper describes a neural-based technique for detecting upright persons in low-resolution beach imagery in order to predict trends of tourist activities at beach sites.

Published: Australasian Joint Conference on Artificial Intelligence

Extensions to Generic Automated Marking Environment: Game-2+
Dr. Steve Green PhD; Chief Research Officer
2009

This paper describes the Generic Automated Marking Environment (GAME-2+), which is the extension of GAME-2 and provides an analysis of its performance in assessing student programming projects.

Published: Proceedings of the Interactive Computer Aided Learning Conference (ICL 2009)

Co-evolutional learning, human / AI behavior

Dynamic Language comprehension

Humanizing AI, biologically inspired

Complex image and landscape analysis, evolutional art

Problem solving - games and conundrums

Problem solving - robotics

Computational Intelligence, Learning  and Problem solving  -  approaches