Our Research

CONNECTED & ADAPTIVE LEARNING INTELLIGENCE SYSTEM

Research Project Scope

From kindergarten to graduate school, one of the key ways artificial intelligence will impact education is through the application of greater levels of individualised learning. Some of this is already happening through growing numbers of adaptive learning programs, games, and software. These systems respond to the needs of the student, putting greater emphasis on certain topics, repeating things that students haven’t mastered, and generally helping students to work at their own pace, whatever that may be. This kind of custom tailored education could be a machine-assisted solution to helping students at different levels work together in one classroom or with teachers facilitating the learning and offering help and support when needed or combined.

Adaptive learning has already had a huge impact on education across the world, and as AI advances in the coming decades adaptive programs like these will likely only improve and expand.

The proposed research project is not aimed at finding different options of how artificial intelligence can replicate the human learning style or replace teachers in class rooms. The aim is to understand how Augmented humanoid learning can make use of human learning behaviours and determine the learning path of a specific learner based on its legacy data of millions of students with the similar persona and behaviour to predict the learning path of new learner and then applying intelligent learning algorithms to produce highly intelligent and tailor made individualised learning curriculum for the learner. This research project aims at developing algorithms, patterns, products and channels for tracking & tagging of student learning style, mental age and learning journey to convert collective learning behaviours into connected learning intelligence (CLI) hence consolidating all of these into a connected, research-based education operating system.

Identification of learning behaviours based on tracking & processing of personal profile, learning issues, mental age, learning styles of group of children (learners). Development of Human Intelligence (HI) based Swarming AI models to convert behavioural learning patterns into connected & collective learning intelligence. Prediction and pattern engine works on persona data sample sets (e.g. age, gender, health condition, disability, family background etc) and internal factors (e.g. learning issues, mental age, learning style) to analyse learning patterns for intelligent curriculum development, self-paced lesson plans and learning assessment.

Problem Statement

The research project will be addressing a wide array of challenges faced by current age educational systems, pedagogy, curriculum design, learning outcomes, instruction relays and pathways and scalability:

Outdated curriculum design and teaching methodology
Traditional education’s insensitivity and in adaptability to learners’ aptitude and needs
Scalability of literacy initiatives is an increasing challenge because of teachers training, availability and quality
Learning quality dependency on trained teachers/student ratio that is increasingly becoming hard to achieve
Traditional education’s inability to identify and address learning challenges
Ineffective literacy improvement efforts without factoring in delicate connection of poverty and child labour in Third World countries
Best instructional contents are not the only ones designed by Subject Specialists. Teachers own contents are product of hands on experience, requirements and understanding of their students. In the absence of a system where they can be added into mainstream curriculum to be shared or re-used, these best practices are not effectively utilised.
Learning environments have not evolved over time to respond to new factors challenging the learning environments and learning behaviours such as digital consumption, pace of learning (fast track), learning practicality (learning concepts in action), learning flexibility (personal pace at desired time).
Rigid time allocation of conventional schooling and tightly integrated role of marginalised child in his/her family financial support system doesn’t encourage parents to let their child into daily schooling
Long hours conventional brick & mortar school commitment prohibit any support in education from his/her labour work/employer
Being part of less privileged family structure, sustaining motivation is a challenge after returning home in the same financially challenged home environment even if child is admitted to normal schooling hours agreeing to leave labour work by parents.
Less privileged child may suffer with increased learning issues or disabilities (learning challenged – on-sit behaviour etc) because of socio-physical factors (in-adequate intake and care during pregnancy and medical facilitation during birth, less hygienic and civilised environment & social activities, inadequate daily nutritional intake & malnutrition growth
Dependancy on teachers, student teacher ratio, infrastructure cost makes educational not scalable and basic literacy for all seems far fetched.

Research Project Phases

IDENTIFICATION

Key Problem Areas: Exploration, identification and verification of key challenges affecting modern educational systems

Infrastructure:

Software Components: AWS SageMaker, AWS Rekognition, AWS Comprehend, Matlab
Hardware Components: AWS EMR Cluster, AWS Redshift Cluster, Elasticsearch Cloud
Data Pipeline Components: Surveys, interviews, on-site observations, literature review, interventions, mock learning environment testing

Download Challanges Brief

ADAPTATION

Virtual Mentoring for Learner: Omnipresent support that integrates user modelling, social simulation and knowledge representation.

Infrastructure:

Software Components: Virtual Class Room Application
Hardware Components: Dev Environment
Data Pipeline Components: Online class sessions

Download Application Brief

Download Architecture Analysis

The Solution

A suite of researched oriented digital learning applications for new disruptive class rooms where child should not struggle to match the pace of teacher for learning, instead each child should have the flexibility of learning at his/her own mental learning pace and getting more learning in same time through fast track learning tools and methodologies, getting assistance in their learning issues and validating the learning outcome they should have with respect to their age and learning levels using their digital addiction as drive for their learning.

Features Highlights

Identify learning issues in normal child e.g dyslexia, dyscalculia, dysgraphia, on-sit behavior, focus issue, eye sight weakness, color blindness, hearing issue, slow mental processing, depth issue
Measure the severity & complexity of learning issues
Validation of learning outcomes with the current learning level to verify the learning issues
Measure the general learning progress with respect to age & learning level (clarity of concepts of learning)
Assist teachers with automated learning material for learning issues & iterative through focused group learning – one group of child with one learning issue – enforced role play fast track learning without the need of individual allocated time for each child
Immersive, fun, interactive fast track learning (3x more learning in same time) with focus on concepts understanding than learning grading through marks
Use the digital addiction as drive for learning in teacher supervised not teacher dependent learning environment
Progress tracking for parents

Mechanism

Child profile data (age etc)
Child issues assessment (manually (phase 1) or automated through activities phase 2)
Learning Activities (to determine learning styles and mental age)
Sequential learning concepts + activities using stack in/stack out, stack up, stack down methodologies
Path identification (mental age, learning style and learning challenges based learning outcome blocks based sequential learning path)
Individualized lesson plan (intelligent curriculum)
Intelligent Lesson plan (N primary concepts memory, N secondary (next), N tertiary (prev) and N auxiliary concept (foundational – future) weak/strong concept identification
Stack in new concept if existing strong
Stack up weak concepts
Stack down strong concepts
Stack out proven concepts based on learning mental age
Learning Outcome Chain: Learning outcome arrangement & linking (primary outcomes & auxiliary outcomes = secondary outcomes, tertiary outcomes)

Benefits

Teaching methodology re-imagined (script based learning, supervised learning, automated learning)
Curriculum design methodology re-imagined (Continuous feedback and assessment loop to improve the curriculum)
Curriculum re-imagined (Flexible (self-paced), fast track, immersive and adaptive and automated sequential learning outcome based curriculum (stack in/stack out – stack up/stack down methodology, learning assessment, motivational and learning issues assessment)
Teaching environment re-imagined
Mobile and open time: learning class with supervised (not instructed) learning sessions) teaching instructions
Student tracking, Learning outcome/concept, activities (script), assessment, pace tracking & adaption, tagging, related apps
Lap Schooling: Parents supervised schooling at home for toddlers (2.5-5) who need closeness of mothers more than a new environment and strangers to adjust in at an age when they are not required to develop any social skills and mother or home training is most important.
Fast track learning (1 hour = 1 conventional school day learning)
Self paced & automated (profile based progress tracking and learning sequence allocation based on daily outcome progress