Course / Reading list
Primary sources
Read the papers, not the summaries.
A verified, filterable list for Lecture 01 and the course. Every arXiv ID and DOI was checked when the lecture was written; venue claims should be re-checked before you quote them.
The reading list
Scope · the Lecture 01 spine
This table is the verified spine of Lecture 01. Every live lecture ends with its own reading list, tied to that lecture's argument. The course-wide library will merge them here as the arc grows; until then, read this table as the foundation, not the whole bibliography.
| Paper | Authors · year | Thread | Why read it | Link |
|---|---|---|---|---|
| Neurosymbolic AI: the 3rd wave | Garcez & Lamb · 2023 | Survey | The framing paper; the “third wave” narrative. | DOI |
| Neurosymbolic AI — Why, What, and How | 2023 | Survey | A compact map of the motivations and methods. | arXiv:2305.00813 |
| From Statistical Relational to Neuro-Symbolic AI | 2020 | Survey | The bridge from probabilistic relational models to NeSy. | arXiv:2003.08316 |
| Neural-Symbolic Learning and Reasoning: A Survey | 2017 | Survey | The older integration literature, still useful. | arXiv:1711.03902 |
| A Survey on Neural-symbolic Learning Systems | 2021 | Survey | A systematic taxonomy of learning systems. | arXiv:2111.08164 |
| Neurosymbolic Reinforcement Learning and Planning: A Survey | 2023 | Survey | The RL/planning angle, closest to embodied applications. | arXiv:2309.01038 |
| Graph Neural Networks Meet Neural-Symbolic Computing | 2020 | Survey | The graph-learning bridge; useful for your background. | arXiv:2003.00330 |
| Verification and Validation of Neurosymbolic AI | 2024 | Survey | What it means to test and assure a NeSy system. | arXiv:2401.03188 |
| Defining neurosymbolic AI | 2025 | Survey | A recent attempt to pin down the definition. | arXiv:2507.11127 |
| Challenges, Opportunities, and Misconceptions in Neurosymbolic AI | 2026 | Survey | The current critical assessment; read after the basics. | arXiv:2611.09356 |
| DeepProbLog: Neural Probabilistic Logic Programming | Manhaeve et al. · 2018 | Interface | Neural predicates inside probabilistic logic; end-to-end training. | arXiv:1805.10872 |
| NeurASP: Embracing Neural Networks into Answer Set Programming | Yang, Ishay & Lee · 2020 | Interface | The Sudoku example; ASP as the symbolic host. | arXiv:2307.07700 |
| Scallop: A Language for Neurosymbolic Programming | 2023 | Interface | A programmable NeSy language with a differentiable engine. | arXiv:2304.04812 |
| ULLER: A Unified Language for Learning and Reasoning | 2024 | Interface | A unified semantics for NeSy languages. | arXiv:2405.00532 |
| Logic Tensor Networks | 2022 | Interface | Fuzzy logic as a differentiable loss. | arXiv:2012.13635 |
| A Semantic Loss Function for Deep Learning with Symbolic Knowledge | 2018 | Interface | The canonical logic-as-loss paper. | arXiv:1711.11157 |
| A-NeSI: A Scalable Approximate Method for Probabilistic Neurosymbolic Inference | 2022 | Interface | Scaling probabilistic NeSy inference. | arXiv:2212.12393 |
| On the Independence Assumption in Neurosymbolic Learning | 2024 | Interface | Why the common independence assumption is dangerous. | arXiv:2404.08458 |
| On the Hardness of Probabilistic Neurosymbolic Learning | 2024 | Interface | The computational limits of the probabilistic approach. | arXiv:2406.04472 |
| Learning Explanatory Rules from Noisy Data | Evans & Grefenstette · 2018 | Interface | Differentiable ILP; learning logic programs from examples. | arXiv:1711.04574 |
| End-to-End Differentiable Proving | Rocktäschel & Riedel · 2017 | Interface | Neural theorem proving; the search interface. | arXiv:1705.11040 |
| DreamCoder: Growing generalizable, interpretable knowledge | Ellis et al. · 2021 | Interface | Library learning and program synthesis. | arXiv:2006.08381 |
| Solving olympiad geometry without human demonstrations | Trinh et al. · 2024 | Interface | AlphaGeometry: a model proposes, a prover checks. | DOI |
| LeanDojo: Theorem Proving with Retrieval-Augmented Language Models | 2023 | Interface | The Lean ecosystem for verified reasoning. | arXiv:2306.15626 |
| Neurosymbolic Conformal Classification | 2024 | Interface | Uncertainty guarantees for NeSy predictions. | arXiv:2409.13585 |
| T-ILR: a Neurosymbolic Integration for LTLf | 2025 | Interface | Temporal logic as a NeSy constraint. | arXiv:2508.15943 |
| Neurosymbolic AI for Reasoning over Knowledge Graphs: A Survey | 2023 | Interface | The data interface: embeddings, constraints, rules. | arXiv:2302.07200 |
| LINC: A Neurosymbolic Approach for Logical Reasoning | 2023 | Language | LLM + first-order logic prover; a clean tool-use baseline. | arXiv:2310.15164 |
| Logic-LM: Empowering LLMs with Symbolic Solvers | Pan et al. · 2023 | Language | Translate, solve, self-refine; the canonical pipeline. | arXiv:2305.12295 |
| LLM+P: Empowering LLMs with Optimal Planning Proficiency | 2023 | Language | Natural language to PDDL to a classical planner. | arXiv:2304.11477 |
| LLMs Can’t Plan, But Can Help Planning in LLM-Modulo Frameworks | Kambhampati et al. · 2024 | Language | The strongest statement of the verifier interface for LLMs. | arXiv:2402.01817 |
| Faith and Fate: Limits of Transformers on Compositionality | Dziri et al. · 2023 | Language | The empirical case for the neural composition gap. | arXiv:2305.18654 |
| Intermediate Languages Matter | 2025 | Language | The formal language you choose changes NeSy LLM reasoning. | arXiv:2509.04083 |
| Neurosymbolic Language Reasoning as Satisfiability Modulo Theory | 2026 | Language | SMT as the interface between language and logic. | arXiv:2602.18095 |
| Do As I Can, Not As I Say: Grounding Language in Robotic Affordances | Ahn et al. · 2022 | Embodiment | SayCan: language grounded in what the robot can do. | arXiv:2204.01691 |
| Code as Policies: Language Model Programs for Embodied Control | Liang et al. · 2022 | Embodiment | Programs as the interface between language and action. | arXiv:2209.07753 |
| AutoTAMP: Autoregressive Task and Motion Planning with LLMs | 2023 | Embodiment | LLMs as translators and checkers for TAMP. | arXiv:2306.06531 |
| ChatHTN: Interleaving Approximate (LLM) and Symbolic HTN Planning | 2025 | Embodiment | HTN planning as the symbolic backbone for an LLM. | arXiv:2505.11814 |
| The Neuro-Symbolic Concept Learner | Mao et al. · 2019 | Embodiment | The hook: perception, programs, and grounded composition. | arXiv:1904.12584 |
| ConceptGraphs: Open-Vocabulary 3D Scene Graphs | 2023 | Embodiment | Scene graphs as the grounding layer for planning. | arXiv:2309.16650 |
| Learning for Long-Horizon Planning via Neuro-Symbolic Abductive Imitation | 2024 | Embodiment | Abduction as the interface between demonstration and plan. | arXiv:2411.18201 |
| Hierarchical Neuro-Symbolic Decision Transformer | 2025 | Embodiment | A recent attempt to put symbols inside a decision transformer. | arXiv:2503.07148 |
| CLEVR: A Diagnostic Dataset for Compositional Language and Visual Reasoning | Johnson et al. · 2017 | Benchmark | The diagnostic benchmark behind the hook; know its limits. | arXiv:1612.06890 |
| FOLIO: Natural Language Reasoning with First-Order Logic | 2022 | Benchmark | A benchmark for faithful logical reasoning in language. | arXiv:2209.00840 |
| LTLZinc: A Benchmarking Framework for Continual Learning and Neuro-Symbolic Temporal Reasoning | 2025 | Benchmark | A benchmark for temporal NeSy reasoning. | arXiv:2507.17482 |
| Embodied Agent Interface: Benchmarking LLMs for Embodied Decision Making | Li et al. · 2024 | Benchmark | The error taxonomy for embodied agents; use it to design evaluations. | arXiv:2410.07166 |
Selection rule
This is a reading list, not a bibliography. It favors papers that made a conceptual contribution or defined a benchmark. Every entry has a stable arXiv ID or DOI; the “why read it” line is a position, not a summary of the abstract.