Toolkit
This page lists the neuro-symbolic AI related repositories being developed at IBM Research. The repositories are categorized in the following eight major categories. There are tags beyond these categories too, these tags show the projects/pipelines where the repository was deployed.
- Logical Neural Network (LNN)
- Natural language processing via reasoning (NLP)
- Knowledge foundation (KF)
- Learning with less (LwL)
- Knowledge augmented sequential decision making (SDM)
- Human in the loop (HIL)
- Datasets and environments (DS)
- Related advances (RA)
Logical Neural Network
FOL-LNN
First-order logic LNN
Logical Optimal Actions (LOA)
LOA is the core of NeSA
Logical Formula Embedder (part of TRAIL)
Logical Formula Embedder (part of TRAIL)
TRAIL
A Deep Reinforcement Learning Approach to First-Order Logic Theorem Proving
Tensor-LNN
GPU scaling of Propositional LNN
NESTA
Learning Symbolic Rules over Abstract Meaning Representations for Textual Reinforcement Learning
NLP via reasoning
Answer Type Prediction
Answer Type Prediction
SLING
Relation linking framework
Sem Rel
Simple transformer-baed Relation linking
GenRL
Relation linking as a generative problem
Knowledge-Enabled Textual-Entailment
Knowledge-Enabled Textual-Entailment
Logic Embeddings
Logic Embeddings
Neural Unification for Logic Reasoning over Language
Neural Unification for Logic Reasoning over Language
TempQA-WD Dataset
Temporal reasoning dataset
AMR Parsing
Abstract Meaning Representation
AMR-to-Logic
Semantic parsing part of the NeSA pipeline to convert natural language text into contextual logic
Knowledge foundation
ERGO
Reasoning enabled graph
Word embeddings from ontologies
Word embeddings from ontologies
HKLIB
HyperKnowledge Library
IT Operations Ontology
IT Operations Ontology
Knowledge-Enabled Textual-Entailment
Knowledge-Enabled Textual-Entailment
Logic Embeddings
Logic Embeddings
TRAIL
A Deep Reinforcement Learning Approach to First-Order Logic Theorem Proving
ULKB
Universal Logic Knowledge Base
AI Descartes
new
Learning with less
Logical Optimal Actions (LOA)
LOA is the core of NeSA
NeSA Demo
Demo for Neuro-Symbolic Agent
NESTA
Learning Symbolic Rules over Abstract Meaning Representations for Textual Reinforcement Learning
AI Descartes
new
Knowledge augmented sequential decision making
E-PDDL
E-PDDL
RLAPSE
Reinforcement Learning with Algorithms from Probabilistic Structure Estimation
TRAIL
A Deep Reinforcement Learning Approach to First-Order Logic Theorem Proving
CREST
Bootstrapped q-learning with context relevant observation pruning
Human in the loop
Ethical AI Platform
Ethical AI Platform
Ethical AI Demo
Ethical AI Demo
NeSA Demo
Demo for Neuro-Symbolic Agent
Datasets and environments
TextWorld Commonsense
Game engine for cleaning room
SCERL
A Text-based Safety Benchmark for Reinforcement Learning Problems
Life Jacket dataset
Life Jacket dataset
Power plant dataset
Power plant dataset
Logical Twins (Textworld-to-PDDLGym)
Wrapper from Textworld to PDDLGym
Related advances
Axiom Verification
Axiom Verification
Binary Matrix Factorization
Binary Matrix Factorization
Ethical AI Platform
Ethical AI Platform
Ethical AI Demo
Ethical AI Demo
Formal ML
Formal ML
GraphSEIR_aPCE
GraphSEIR_aPCE
Designing new fair voting rules
Designing new fair voting rules
Policy Gradient Algorithm for Learning to Learn in Multiagent RL
Policy Gradient Algorithm for Learning to Learn in Multiagent RL
Framework for collecting moral preference data
Framework for collecting moral preference data
MER
Meta-Experience Replay
Online Alternating Minimization
Online Alternating Minimization
Unsupervised Learning of Graph Hierarchical Abstractions
Unsupervised Learning of Graph Hierarchical Abstractions with Differentiable Coarsening and Optimal Transport
Sobolev Independence Criterion
Sobolev Independence Criterion
Persistence Homology for Link Prediction
Persistence Homology for Link Prediction
TM-GCN
TM-GCN
Ethical AI
Axiom Verification
Axiom Verification
Ethical AI Platform
Ethical AI Platform
Ethical AI Demo
Ethical AI Demo
Designing new fair voting rules
Designing new fair voting rules
Framework for collecting moral preference data
Framework for collecting moral preference data
Life Jacket dataset
Life Jacket dataset
Power plant dataset
Power plant dataset
NSQA
Answer Type Prediction
Answer Type Prediction
ERGO
Reasoning enabled graph
SLING
Relation linking framework
Sem Rel
Simple transformer-baed Relation linking
GenRL
Relation linking as a generative problem
Logic Embeddings
Logic Embeddings
TempQA-WD Dataset
Temporal reasoning dataset
AMR Parsing
Abstract Meaning Representation
NeSA
Logical Optimal Actions (LOA)
LOA is the core of NeSA
NeSA Demo
Demo for Neuro-Symbolic Agent
NESTA
Learning Symbolic Rules over Abstract Meaning Representations for Textual Reinforcement Learning
AMR-to-Logic
Semantic parsing part of the NeSA pipeline to convert natural language text into contextual logic
CREST
Bootstrapped q-learning with context relevant observation pruning
Logical Twins (Textworld-to-PDDLGym)
Wrapper from Textworld to PDDLGym