M32.6 CONNECT THE MECHANISM
Connect learned representations with explicit reasoning constraints
A network learns to read handwritten digits without ever seeing a digit label, only their sums. Logic supplies the missing step and sends gradients back through it.
LESSON OVERVIEW13 min lesson
Lesson overview
A network learns to read handwritten digits without ever seeing a digit label, only their sums. Logic supplies the missing step and sends gradients back through it.
What you’ll explore
- Neuro-symbolic methods combine neural learning with symbolic structures or constraints; soft penalties, differentiable reasoning, and hard verification provide different guarantees and failure modes.
GO TO THE SOURCE
Original explanations, connected to the research.
DeepProbLog: Neural Probabilistic Logic Programming (Manhaeve et al., 2018)A Semantic Loss Function for Deep Learning with Symbolic Knowledge (Xu et al., 2018)Logic Tensor Networks (Serafini & d'Avila Garcez, 2016)DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning (Ellis et al., 2020)Solving olympiad geometry without human demonstrations (AlphaGeometry; Trinh et al., 2024)Wu's Method can Boost Symbolic AI to Rival Silver Medalists and AlphaGeometry to Outperform Gold Medalists at IMO Geometry (Sinha et al., 2024)Suggest a correction
A precise note can make an explanation better.
Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.