> For the complete documentation index, see [llms.txt](https://neurosymbolicai.gitbook.io/neuro-symbolic-ai-in-practice/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://neurosymbolicai.gitbook.io/neuro-symbolic-ai-in-practice/part-iii-core-approaches/chapter-4/4-2-neural-helps-symbolic.md). # 4.2 Neural Helps Symbolic > *"The symbolic framework retains its formal guarantees — soundness, decidability, interpretability — while the neural component dramatically reduces computational cost."* *** In this paradigm, neural networks act as subroutines to **accelerate, approximate, or replace** computationally expensive components within an otherwise traditional symbolic framework. The symbolic framework retains its formal guarantees — soundness, decidability, interpretability — while the neural component dramatically reduces computational cost. The core intuition: symbolic reasoning can be extraordinarily expensive. A classical planner searching an exponential state space, a formal theorem prover exploring a combinatorial proof space, a constraint solver examining millions of variable assignments — all computationally hard problems where neural approximations can provide order-of-magnitude speedups at the cost of occasionally imprecise guidance. Because the symbolic framework verifies correctness, imprecise neural guidance does not compromise output quality — it only affects efficiency. > \[!IMPORTANT] The key safety property of this paradigm: the symbolic framework is the source of truth. A neural subroutine that gives bad guidance causes the symbolic system to run slower (more search) or produce suboptimal solutions — but never to produce incorrect or unsafe outputs. This is a fundamentally different risk profile from neural-only systems, where errors in the model directly produce errors in the output. ## Sections in This Chapter * [4.2.1 Differentiable Reasoning](/neuro-symbolic-ai-in-practice/part-iii-core-approaches/chapter-4/4-2-neural-helps-symbolic/4-2-differentiable-reasoning.md) — MLN, LTN, DeepProbLog, Scallop, ∂ILP, FlashFill, DreamCoder, Popper, ILASP * [4.2.2 Neural Subroutines in Symbolic Systems](/neuro-symbolic-ai-in-practice/part-iii-core-approaches/chapter-4/4-2-neural-helps-symbolic/4-2-neural-subroutines.md) — Neural heuristics, MCTS/AlphaProof, HTPS, Tactician, LATS, Neural Combinatorial Optimization, GFlowNets, NeuroSAT, SATNet * [4.2.3 Answer Set Programming](/neuro-symbolic-ai-in-practice/part-iii-core-approaches/chapter-4/4-2-neural-helps-symbolic/4-2-asp.md) — NeurASP, DeepStochLog, SLASH, comparison table, Summary, Exercises, References --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://neurosymbolicai.gitbook.io/neuro-symbolic-ai-in-practice/part-iii-core-approaches/chapter-4/4-2-neural-helps-symbolic.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.