Neuro-symbolic AI is a type of
artificial intelligence that integrates
neural and
symbolic AI architectures to address the weaknesses of each, providing a robust AI capable of
reasoning, learning, and
cognitive modeling. As argued by
Leslie Valiant[1] and others,[2][3] the effective construction of rich computational
cognitive models demands the combination of
symbolic reasoning and efficient
machine learning.
Gary Marcus, argued, "We cannot construct rich
cognitive models in an adequate, automated way without the triumvirate of hybrid architecture, rich prior knowledge, and sophisticated techniques for reasoning."[4] Further, "To build a robust, knowledge-driven approach to AI we must have the machinery of symbol manipulation in our toolkit. Too much useful knowledge is abstract to proceed without tools that represent and manipulate abstraction, and to date, the only known machinery that can manipulate such abstract knowledge reliably is the apparatus of symbol manipulation."[5]
Henry Kautz,[6]Francesca Rossi,[7] and
Bart Selman[8] also argued for a synthesis. Their arguments attempt to address the two kinds of thinking, as discussed in
Daniel Kahneman's book
Thinking Fast and Slow. It describes cognition as encompassing two components: System 1 is fast, reflexive, intuitive, and unconscious. System 2 is slower, step-by-step, and explicit. System 1 is used for
pattern recognition. System 2 handles planning, deduction, and deliberative thinking. In this view,
deep learning best handles the first kind of cognition while symbolic reasoning best handles the second kind. Both are needed for a robust, reliable AI that can learn, reason, and interact with humans to accept advice and answer questions. Such dual-process models with explicit references to the two contrasting systems have been worked on since the 1990s, both in AI and in Cognitive Science, by multiple researchers.[9]
Approaches
Approaches for integration are diverse.
Henry Kautz's taxonomy of neuro-symbolic architectures,[10] along with some examples, follows:
Symbolic Neural symbolic is the current approach of many neural models in
natural language processing, where words or subword tokens are the ultimate input and output of
large language models. Examples include
BERT, RoBERTa, and
GPT-3.
Symbolic[Neural] is exemplified by
AlphaGo, where symbolic techniques are used to invoke neural techniques. In this case, the symbolic approach is
Monte Carlo tree search and the neural techniques learn how to evaluate game positions.
Neural | Symbolic uses a neural architecture to interpret perceptual data as symbols and relationships that are reasoned about symbolically. Neural-Concept Learner[11] is an example.
Neural: Symbolic → Neural relies on symbolic reasoning to generate or label
training data that is subsequently learned by a deep learning model, e.g., to train a neural model for symbolic computation by using a
Macsyma-like
symbolic mathematics system to create or label examples.
Neural_{Symbolic} uses a
neural net that is generated from symbolic rules. An example is the Neural Theorem Prover,[12] which constructs a neural network from an
AND-OR proof tree generated from knowledge base rules and terms. Logic Tensor Networks[13] also fall into this category.
Neural[Symbolic] allows a neural model to directly call a symbolic reasoning engine, e.g., to perform an action or evaluate a state. An example would be
ChatGPT using a
plugin to query
Wolfram Alpha.
These categories are not exhaustive, as they do not consider multi-agent systems. In 2005, Bader and
Hitzler presented a more fine-grained categorization that considered, e.g., whether the use of symbols included logic and if it did, whether the logic was
propositional or first-order logic.[14] The 2005 categorization and Kautz's taxonomy above are compared and contrasted in a 2021 article.[10] Recently,
Sepp Hochreiter argued that
Graph Neural Networks "...are the predominant models of neural-symbolic computing"[15] since "[t]hey describe the properties of molecules, simulate social networks, or predict future states in physical and engineering applications with particle-particle interactions."[16]
Artificial general intelligence
Gary Marcus argues that "...hybrid architectures that combine learning and symbol manipulation are necessary for robust intelligence, but not sufficient",[17] and that there are
...four cognitive prerequisites for building robust artificial intelligence:
hybrid architectures that combine large-scale learning with the representational and computational powers of symbol manipulation,
large-scale knowledge bases—likely leveraging innate frameworks—that incorporate symbolic knowledge along with other forms of knowledge,
reasoning mechanisms capable of leveraging those knowledge bases in tractable ways, and
A series of workshops on neuro-symbolic AI has been held annually since 2005 Neuro-Symbolic Artificial Intelligence.[23] In the early 1990s, an initial set of workshops on this topic were organized.[19]
What is the best way to integrate neural and symbolic architectures?
How should symbolic structures be represented within neural networks and extracted from them?
How should common-sense knowledge be learned and reasoned about?
How can abstract knowledge that is hard to encode logically be handled?
Implementations
Implementations of neuro-symbolic approaches include:
AllegroGraph: an integrated Knowledge Graph based platform for neuro-symbolic application development.[25][26][27]
Scallop: a language based on
Datalog that supports differentiable logical and relational reasoning. Scallop can be integrated in
Python and with a
PyTorch learning module.[28]
Logic Tensor Networks: encode logical formulas as neural networks and simultaneously learn term encodings, term weights, and formula weights.
DeepProbLog: combines neural networks with the probabilistic reasoning of
ProbLog.
SymbolicAI: a compositional differentiable programming library.
Explainable Neural Networks (XNNs): combine neural networks with symbolic
hypergraphs and trained using a mixture of backpropagation and symbolic learning called induction.[29]
^Serafini, Luciano; Garcez, Artur d'Avila (2016). "Logic Tensor Networks: Deep Learning and Logical Reasoning from Data and Knowledge".
arXiv:1606.04422 [
cs.AI].
Bader, Sebastian;
Hitzler, Pascal (2005-11-10). "Dimensions of Neural-symbolic Integration – A Structured Survey".
arXiv:cs/0511042.
Garcez, Artur S. d'Avila; Broda, Krysia; Gabbay, Dov M.; Gabbay (2002). Neural-Symbolic Learning Systems: Foundations and Applications. Springer Science & Business Media.
ISBN978-1-85233-512-0.
Garcez, Artur d'Avila; Gori, Marco; Lamb, Luis C.; Serafini, Luciano; Spranger, Michael; Tran, Son N. (2019). "Neural-Symbolic Computing: An Effective Methodology for Principled Integration of Machine Learning and Reasoning".
arXiv:1905.06088 [
cs.AI].
Garcez, Artur d'Avila; Lamb, Luis C. (2020). "Neurosymbolic AI: The 3rd Wave".
arXiv:2012.05876 [
cs.AI].
Hochreiter, Sepp. "Toward a Broad AI." Commun. ACM 65(4): 56–57 (2022).
Toward a broad AI
Honavar, Vasant (1995). Symbolic Artificial Intelligence and Numeric Artificial Neural Networks: Towards a Resolution of the Dichotomy. The Springer International Series In Engineering and Computer Science. Springer US. pp. 351–388.
doi:
10.1007/978-0-585-29599-2_11.
Serafini, Luciano; Garcez, Artur d'Avila (2016-07-07). "Logic Tensor Networks: Deep Learning and Logical Reasoning from Data and Knowledge".
arXiv:1606.04422 [
cs.AI].
Sun, Ron (1995). "Robust reasoning: Integrating rule-based and similarity-based reasoning". Artificial Intelligence. 75 (2): 241–296.
doi:
10.1016/0004-3702(94)00028-Y.
Sun, Ron; Bookman, Lawrence (1994). Computational Architectures Integrating Neural and Symbolic Processes. Kluwer.