AI Apparatus Autonomous Knowledge Expansion via Pattern Conversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional artificial intelligence techniques based on expert systems and neural networks face challenges in expressing human thinking patterns, handling illogical items, and ensuring knowledge system consistency with reality, while also being prone to slow convergence and black box processes.
Innovation Solution
The development of an artificial intelligence apparatus that converts human thoughts expressed in words, numbers, and symbols into patterns, allowing for autonomous construction of a knowledge system, activation of relevant patterns, and execution of processes, ensuring transparency and consistency with human rules and morals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If expert systems use rule-based reasoning engines, then logical processing capability is improved, but inability to handle illogical items and conflict between rules worsens
Solution Approach 1:
The patent introduces a knowledge base constructed from diverse information sources (textbooks, encyclopedias, news, etc.) as an intermediary between the reasoning engine and reality. This knowledge base serves as a mediator that provides contextual understanding and factual grounding, enabling the system to handle illogical items by referencing external knowledge rather than relying solely on internal rule consistency.
Solution Approach 2:
The patent replaces the rigid mechanical rule-based system with a hybrid approach that incorporates neural network components and natural language processing. This substitution allows the system to interpret and reason about illogical items by understanding their semantic meaning and contextual relevance, rather than strictly applying predetermined rules.
2Reliability
If neural networks use back propagation method, then learning capability is improved, but convergence speed becomes very slow and convergence is not guaranteed
Solution Approach 1:
The patent pre-constructs a comprehensive knowledge base containing structured information from multiple sources before the learning process begins. This preliminary action provides the neural network with initial knowledge and contextual understanding, reducing the time required for convergence by eliminating the need to learn basic facts and relationships from scratch during the training phase.
Solution Approach 2:
The patent segments the learning process into distinct phases: (1) knowledge base construction from external sources, (2) neural network training on specific tasks, and (3) reasoning and problem-solving. This segmentation allows the system to leverage pre-processed knowledge for rapid convergence on specific tasks without requiring extensive retraining on general knowledge.
3Reliability
If neural networks use back propagation method, then learning capability is improved, but it leads to local optimum points instead of broad-based optimized solution
Solution Approach 1:
The patent implements multiple feedback mechanisms: (1) The knowledge base provides factual feedback to verify the correctness of learned representations, (2) The reasoning engine provides logical feedback to validate inference paths, and (3) External evaluation metrics provide performance feedback to guide continued optimization. This multi-layered feedback system helps escape local optima by identifying and correcting suboptimal solutions through logical and factual verification.
Solution Approach 2:
The patent merges neural network learning with symbolic reasoning and knowledge base verification. This combination allows the system to evaluate solutions not only based on statistical performance but also on logical consistency and factual accuracy, providing a more comprehensive optimization criterion that avoids local optima and achieves broader, more robust solutions.
4Extent of automation
If conventional AI processes are used, then automation is improved, but the process becomes a black box and validity is not clear
Solution Approach 1:
The patent segments the AI system into distinct, interpretable components: knowledge base construction, information processing, reasoning engine, and output generation. Each component operates with clear inputs and outputs that can be traced and understood, maintaining transparency while achieving automation. The reasoning engine specifically provides step-by-step logical derivations that make the decision-making process visible and validatable.
Solution Approach 2:
The patent introduces the knowledge base and reasoning engine as intermediaries between the neural network and the final output. These intermediaries act as interpreters that translate neural network representations into human-understandable logical reasoning steps, maintaining transparency and validity while preserving autonomous processing capability. The reasoning engine specifically provides explainable inference paths that reveal the validity of conclusions.
Data Source
AI summary
An artificial intelligence apparatus includes an input processor configured to convert input information to patterns, an analyzer configured to analyze the input information, a recorder configured to record the information, a controller configured to perform at least one of a development of a process according to a type of a sentences and an intention, a search for information and a logic development to solve a problem, an execution of a process and activating a program, a generalization of information and a procedure, an update to a better knowledge and a logic, a search and an arrangement of information about an interesting field and an item, recording and updating of information, connective relations and relationship, and a transition control between information to a goal and an output processor configured to convert the patterns to information or control signals.