Intelligent Learning Software for Children with Autism using Machine Learning Agents
Project Title: Intelligent Learning Software for Children with Autism using Machine Learning Agents
Primary Supervisor: Dr Charaka Palansuriya
Additional Supervisor(s): Dr Aurora Constantin
Autism Spectrum Disorder (ASD) describes a range of life-long neurodevelopmental conditions which is usually characterised by impairments in three core domains: social interaction, communication and imagination (known as triad of impairments). Today’s prevalence of ASD in the UK is more than 1 in 100 children.
A high percent of the children with ASD have co-morbidities, including learning disability. Although no treatment exists for ASD, research revealed that children with ASD can benefit from the educational interventions. Particularly, technology-based interventions are very promising as children with ASD have affinity toward technology. In addition, technology has a series of advantages for these children. For example, technology acts an interface between individuals with ASD and other people and that creates emotional and social distancing which is likely to diminish the anxiety. Another advantage is that technology is well-placed to customise the interventions to the child’s particular needs and interests. For instance, children with ASD may have particular likes or dislikes for various sounds, colours and other sensory outputs produce by mobile devices, or have particular sometimes peculiar interests (e.g. in spiders). Moreover, the autistic population is known as being extremely heterogenous. Thus, children with ASD have highly varying learning abilities. While children with Asperger Syndrome (one of the conditions included in ASD) have normal language and above average intelligence, other autistic children are completely nonverbal and present severe intellectual disability. Consequently, these children require personalised one-to-one guidance to learn both life-long skills and academics skills. Training adequate number of teachers and assistants to be specialised in providing this type of guidance is challenging to say the least.
While technology could provide support to children with ASD, given the heterogeneity of the population, it is unlikely that a software that works a predefined way could work well for all these children. However, recent research has shown that Reinforcement Learning based Machine Learning (ML) agents could provide sufficient intelligence and dynamic adjustability to ensure personalised support and guidance. Since such intelligent learning and dynamically adjusting software could be replicated as much as needed and could be easily deployed on mobile platforms like tablets, this could provide effective assistance to teachers and parents who are involved in teaching children with ASD.
The research work will include finding out
- What are the appropriate Machine Learning techniques to use for building intelligence into game-like teaching educational tool for children with ASD. Reinforcement Learning, Imitation Learning and combinations of the two could be used to build such an educational tool. Other supervised or unsupervised learning techniques could also be explored.
- How to speed up training the ML agents that will be embedded in the software. The training of ML agents will be explored in both single machines (e.g., developer laptops) as well as HPC computing platforms managed by EPCC.
- How effective is using game development technologies like Unity, Unreal Engine or OpenAI Gym to develop intelligent learning software for children with ASD? Many of these game development technologies now include machine learning capabilities to adapt the software dynamically to the suitability of the user – for example, adapting the difficulty level and how the guided teaching is delivered to a child with ASD.
- Appropriate way to perform and evaluate usability of the learning technologies developed.
A UK 2:1 honours degree, or its international equivalent, in a relevant subject such as computer science and informatics, physics, mathematics, engineering, biology, chemistry and geosciences.
You must be a competent programmer in at least one of C, C++, C#, Python, Fortran, or Java and should be familiar with mathematical concepts such as algebra, linear algebra and probability and statistics.
Further information and requirements available on University Degree Finder: https://www.ed.ac.uk/studying/postgraduate/degrees/index.php?r=site/view&edition=2020&id=855
Student Recommended/Desirable Skills and Experience
(Anything which a student doesn’t *need* to start the project, but which would make a student a stand-out candidate).
Knowledge on machine learning techniques such as Reinforcement Learning, Supervised and/or Unsupervised learning is a desired skill. It is also desirable to have some experience with developing visual game like environments – particularly using Unity (https://unity.com/) or Unreal Engine (https://www.unrealengine.com/en-US/). Experience in developing game assets using GIMP, Photoshop, Blender, etc. will be useful.
- Unity Machine Learning Agents, https://blogs.unity3d.com/2017/09/19/introducing-unity-machine-learning-agents/
- Unreal Engine Neurostudio, https://unrealengine.com/marketplace/en-US/product/neurostudio-self-learning-ai
- Open AI Gym, https://gym.openai.com/
- Training your agents 7 times faster with ML-Agents, https://blogs.unity3d.com/2019/11/11/training-your-agents-7-times-faster-with-ml-agents/
- Designing Serious Game Interventions for Individuals with Autism, https://link.springer.com/article/10.1007/s10803-014-2333-1