10. ML IN STARTUPS
Motion AIHeyzapThread Yummly
16.32M$
8M$
45M$ exit
1M$ 23M$
Knewton
157M$
11. ML IN EDUCATION
• Applications can be divided into in-class (offline)
and online
12. IN CLASSROM / OFFLINE
• Difficult to collect; data classrooms have limited
number of student
• ML needs data
• Examples: Matching teachers and schools, Grading
systems
13. ONLINE EDUCATION
• Easy to collect data
• Easy to apply results of algorithms
• Much more applications areas
14. ML IN OTSIMO
• Otsimo is a small Startup
• Build product with minimum afford
• Use existing solutions and API’s
15. HOW TO LEARN
Intro to Machine
Learning
Andrew Ng: Machine
Learning
Github Open
Source Projects
github.com/tensorflow/models
and many more
16. HOW TO LEARN
Intro to Machine
Learning
Andrew Ng: Machine
Learning
Github Open
Source Projects
github.com/tensorflow/models
and many more
25. ADAPTIVE LEARNING
• There are different approaches to adaptive learning
• Most common is Bayesian KnowledgeTracing
• Deep KnowledgeTracing
26. DEEP KNOWLEDGE TRACING
• From Stanford University researchers
• Horizontal Axis is timeseries, Vertical Axis is features
• Recurrent Neural Networks to model student learning
27. ITEM RESPONSE THEORY
• Used by knewton
• Knewton claims that its better than DKT
• A single number, called the proficiency or ability,
represents a student’s knowledge state during the
course of completing several assessments