This repository is dedicated to exploring and documenting various neurosymbolic AI architectures. The goal is to understand how neural networks and symbolic reasoning can be combined to create intelligent systems. 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. It basically combines neural networks with symbolic reasoning to create powerful systems.
Neurosymbolic AI merges the learning capabilities of neural networks with the logic and reasoning abilities of symbolic systems. This approach aims to improve both the performance and interpretability of AI models. This repository provides a collection of architectures, research papers, and explanations to help you get started with neurosymbolic AI.
- A neural network model is trained using the dataset to learn patterns and relationships between ingredients and flavor profiles.
- The model is designed to predict the best mocktail recipes based on user preferences.
- The system presents predefined flavor combination options to the user.
- The symbolic logic is used to understand the effects of combining different ingredients. This logic ensures that the mocktail prediction doesn't include a combination of ingredients that will not go together.
In this architecture, the model uses symbolic logic to filter out predictions generated by the model and present the best ones to the user. The symbolic component ensures that the recommendations align with predefined flavor profiles and user preferences, while the neural network leverages data to optimize the predictions. This combination results in a robust system that provides personalized and refined recommendations to the user.
This section explores architectures that tightly integrate neural networks with symbolic reasoning systems. This includes models that use neural networks to process raw data and symbolic systems to reason and make decisions based on high-level abstractions.
Neuro-symbolic programming combines neural networks with programming languages, enabling symbolic logic within neural architectures. This section covers models that incorporate programming constructs to enhance neural network capabilities.
Hybrid approaches aim to leverage the strengths of both neural networks and symbolic reasoning by combining them in a modular fashion. This section includes models that use neural networks for perception and symbolic reasoning for decision-making and problem-solving.
The papers directory contains summaries and links to important research papers in neurosymbolic AI. These papers provide foundational knowledge and recent advancements in the field.
- Neurosymbolic AI: the 3rd wave link to article