22 questions · 9 almanac · 5 findings

Machine Learning

Algorithms that learn patterns from data.

Almanac Teaching Machines to Ask What If
  • How do Bayesian networks let AI systems represent and update uncertain knowledge?
  • What is do-calculus, and how does it distinguish correlation from causation?
  • Why does the ability to reason about interventions matter for fields like epidemiology and economics?
Almanac A Vocabulary for Learning by Reward
  • How did Barto and Sutton's 1998 textbook turn scattered ideas about reward-based learning into a unified discipline?
  • What are temporal-difference learning, Q-learning, and policy gradient methods, and why did they become core algorithms of reinforcement learning?
  • How do systems like AlphaGo and superhuman game-playing agents trace their lineage back to Barto and Sutton's framework?
Almanac The Vocabulary Reward Learning Now Speaks
  • What unifying framework did Barto and Sutton bring to reinforcement learning that had previously been a collection of ad hoc techniques?
  • What is the actor-critic architecture, and how does it fit into the value-function and policy vocabulary Barto and Sutton established?
  • Which modern AI systems trace their lineage back to Barto and Sutton's foundational reinforcement learning work?
Finding CLARE: Cognitive Load Assessment in REaltime with Multimodal Data
  • Which sensor combination and model architecture performed best under 10-fold cross-validation versus leave-one-subject-out (LOSO) evaluation, and what does that difference imply about generalizing cognitive load classifiers to new individuals?
  • Why do CNN-based deep learning models outperform other approaches for binary cognitive load classification on this multimodal dataset?
Finding Leveraging Deep Reinforcement Learning for Metacognitive Interventions across Intelligent Tutoring Systems
  • How does a deep reinforcement learning agent that adapts to a student's changing metacognitive level compare to a static classifier-based approach for teaching strategy-switching in intelligent tutoring systems?
  • What evidence shows that DRL-based metacognitive interventions in one tutoring domain (a logic tutor) transfer to prepare students for future learning in an unrelated domain (a probability tutor)?