Dream to Learn is shutting down...

We are very sorry to say that Dream to Learn will be shutting down as of December 28th, 2019. If you have content that you wish to keep, you should make a copy of it before that date.


Improve Learning Outcomes for Spectrum Edge Learners

POSTED IN: Cognitive Wingman

Identifying Knowledge Milestones to Improve Learning Outcomes for Spectrum Edge Learners




The current education system is primarily focused on the approximately 85-90% of students who are typical learners, and who respond well to typical curricula.  Many students do not fit into this category – the “spectrum learners”; they include:


· Variety of students on the Autism Spectrum (~2%) who need a more nuanced approach

· Gifted students (~6%) bored under-challenged, leading to a failure to thrive to potential

· Other Students with atypical learning styles, failing to thrive with conventional curriculum


The solution helps identify critical path knowledge milestones and optimal paths to improve learning outcomes for spectrum learners.  For this summary, we focus autistic learners.


Key Elements of Solution:

1. Augments known educational best practices (learning outcomes, learning pathways) with insights gained from ML/DL analysis combined with traditional data science

2. Leverages dynamic segmentation and clustering (cohorts, archetypes)

3. Uses deep learning to surface key features in natural language and knowledge set / ontology

4. Uses Machine Learning (e.g. Random Forest) to surface key features in learning path

5. Applies Natural Language Understanding for signal extraction from students & teachers

6. Interacts – Provides natural language and visual interactions to exchange information, including but not limited to Augmented Reality, Virtual Reality and Digital Humans.

7. Learns.  Evolves with the content, learners and teachers to improve over time


Using the solution and elements above, along with established teaching methods – we can compose a dynamic Sensemaking System data driven analysis of what factors lead to successful learning outcomes.  


Proposal: By identifying key components in learning journeys (using prior knowledge of learning outcomes, and other learning journeys) the system will permit Teachers, Parents and Students to understand and address knowledge and learning gaps most likely to increase the likelihood of success and learning outcomes.


Hypothesis:  With the system described we believe we can create an improvement (KPI%-TBD) learning outcomes for students with autism, when compared to those not leveraging solution





The advantage of this system is that it can take a holistic view of thousands of learning journeys, and identify the factors that are impactful to positive/negative learning outcomes, and then (a) surface information for teachers and/or (b) dynamically adjust content or user experience to help the student accomplish the critical path items.



a) Ontology and Content Extraction

a. Leveraging NLU and Data Science methods to tag and cluster a wide variety of Knowledge, Ideas, Concepts and teaching delivery methods, not previously mapped, nor available to analytics. (Data and Corpus ingestion and curation)

b. Modeling and Clustering

c. Surfacing content, components and ideas that are candidates for benefitting some cohorts of learnings

d. user interface / visualization

b) Modeling of Learning Flows

a. mapping covering majority of types;

b. user interface / visualization – possibly leveraging cognitive/spatial for students and teachers – visualizing gaps; alternate pathways

c) Analytics

a. Sequence Analysis / Flag Detection - potentially informative flags, signals and insights to teachers.  For example, if another SEQUENCE of learning or method proved to be 3X more effective for learner type ABC – system could suggest exploring alternate method, content or sequence






4. Description


STEP 1:  Analyze Students and Cluster into multiple Cohorts based on all available data



STEP 2:  Analyze Content of Learning, Curricula, Conversations, Teacher/Therapist Notes




STEP 3:  Produce a Simplified Knowledge Journey (by cohort), segment by outcomes


STEP 4:  Manually (initially) develop Preferred Path and content – best fit for cohort


STEP N:  On wide deployment (Phase 3+) – Consider Digital Humans as mechanism to (a) improve information coming from Student; Teacher; Therapists; and (b) as alternate-path cognitive assistant 
(Image source: soul machines web site, just used as a digital human example - see also faceme or unity models)



SUMMARY:  Two phase process – Phase 1 is Data Analysis and Developing Test; Phase 2 POC





a) DEEP LEARNING / MACHINE LEARNING – automated system will continually re-evaluate based on features (students, cohorts and content/knowledge) to flag essential components and/or engage student or teacher to improve on it

b) EMPATHY / FRAME OF REFERENCE (POV) – modify curricula to reflect the differing perspectives of learners – use ML/DL feature extraction and prediction to elevate importance of most impactful knowledge milestones depending on the cohort and success factors for similar learners;

c) PIVOTAL IDEAS / FLAGS – discovering high-power concepts, ideas, methods and surfacing for wider community to assess, and/or test – methods of automation for validation

d) VISUAL GATES REPRESENTED BY DIGITAL HUMANS (narrow domain) – e.g. UX interaction in Augmented Reality with a cognitive wingman – verbal conversation to solicit information; signal and help in mastery of concept using different methods (at scale) to understand best fit for Cohort/type








Interested in more content by this author?

About the Author

Ryan Anderson

Ryan Anderson

Hi! I like to play with data, analytics and hack around with robots and gadgets in my garage. Lately I've been learning about machine learning.

About this blog

Cognitive Wingman - Never Walk Alone

Created: December 20, 2016


This Blog Appears in

Up Next