AI Basics
Basic concepts of AI and machine learning applied to manufacturing
Machine Learning
Machine learning is a core AI technology that learns patterns from data to automate various manufacturing decisions including quality inspection, predictive maintenance, and process optimization.
Deep Learning
Deep learning is a technology that learns complex patterns using multi-layered neural networks, applied in manufacturing for vision inspection, predictive maintenance, and process optimization to achieve quality improvement and cost reduction.
Artificial Neural Network (ANN)
Artificial Neural Network is a machine learning model inspired by brain neurons, widely used in manufacturing for automated quality inspection, predictive maintenance, and process optimization.
Natural Language Processing (NLP)
Natural Language Processing (NLP) is an AI technology enabling computers to understand and process human language, revolutionizing communication between operators and systems in manufacturing through voice commands, automated document analysis, and knowledge search.
Large Language Model (LLM)
Large Language Models (LLMs) are language generation models trained on vast text data, utilized in manufacturing for work instruction automation, quality analysis, and technical support chatbots.
RAG (Retrieval-Augmented Generation)
RAG is a technique enabling LLMs to retrieve external documents before generating responses, providing accurate answers in manufacturing by referencing equipment manuals and quality data.
Computer Vision
Computer vision is an AI technology that analyzes digital images to extract information for decision-making, playing a critical role in manufacturing for quality inspection automation, process monitoring, and robot vision applications.
Reinforcement Learning
Reinforcement learning is an AI technique that learns action strategies to maximize rewards through trial and error, applied in manufacturing for robot control, production scheduling, and process optimization.
Transfer Learning
Transfer learning is an efficient AI technique that leverages pre-trained model knowledge to rapidly deploy defect inspection and predictive maintenance systems in manufacturing with minimal data.
Anomaly Detection
Anomaly detection is a technique that identifies rare items deviating significantly from normal data patterns, enabling early detection of equipment failures, product defects, and process abnormalities in manufacturing for preventive action.
Classification (Statistical Classification)
Classification is a statistical machine learning technique that assigns input data to predefined categories, widely used in manufacturing for quality inspection automation, equipment fault diagnosis, and process state monitoring.
Regression Analysis
Regression analysis is a statistical technique for mathematically estimating relationships between process variables and quality indicators, fundamentally used in manufacturing for quality prediction and process optimization.
Generative AI
Generative AI automatically creates text, images, code, and other content based on learned data patterns, applied in manufacturing for design automation, technical documentation, and code generation.
Transformer (Deep Learning Architecture)
Transformer is a deep learning architecture based on multi-head attention mechanism, applied in manufacturing for sensor data analysis, quality inspection, and production optimization.
Prompt Engineering
Prompt engineering is the technique of structuring inputs to obtain desired outputs from generative AI, applied in manufacturing for quality analysis, equipment diagnostics, and process optimization.
Fine-tuning
Fine-tuning is a transfer learning technique that adapts pre-trained AI models to specific manufacturing environments and data, achieving high accuracy with minimal data and time investment.
GPT (Generative Pre-trained Transformer)
GPT is a transformer-based generative language model used in manufacturing for technical documentation automation, operator support chatbots, and quality analysis, reducing repetitive tasks and improving knowledge accessibility.
Diffusion Model
Diffusion models are generative AI that learn noise addition and restoration processes to create new data, used in manufacturing for scarce defect data augmentation, design variant generation, and sensor data completion.
AI Agent
AI Agent is a system that autonomously makes decisions and takes actions without continuous human supervision, executing real-time equipment control, quality management, and production planning in manufacturing environments.
Multimodal AI
Multimodal AI integrates and analyzes multiple data types including images, sensors, audio, and text to discover patterns in manufacturing defect inspection, predictive maintenance, and safety management that are difficult to find with single data sources.
Edge AI
Edge AI is a technology that performs AI computations directly on manufacturing equipment, enabling real-time quality inspection, predictive maintenance, and process optimization.
Federated Learning
Federated learning is a technique where multiple manufacturing sites collaboratively train AI models without sharing their data externally, enabling multi-factory quality prediction and supply chain collaborative AI while maintaining data security.
CNN (Convolutional Neural Network)
CNN is a deep learning neural network that automatically learns image patterns, serving as a core technology for automated defect inspection and real-time quality monitoring in manufacturing.
RNN (Recurrent Neural Network)
RNN is a neural network that learns sequential patterns in time-series data such as sensor readings and process variables, utilized in manufacturing for predictive maintenance, quality forecasting, and demand prediction.
Supervised Learning
Supervised learning is an AI method that learns from labeled data, critically utilized in manufacturing for quality inspection automation, predictive maintenance, and process optimization.
Unsupervised Learning
Unsupervised learning enables AI to discover patterns from unlabeled data, effectively used in manufacturing for equipment anomaly detection and quality clustering without manual labeling.
Data Labeling
Data labeling is the process of attaching ground truth tags to raw data to create AI training datasets, serving as a critical step in building defect inspection, predictive maintenance, and process optimization AI in manufacturing.
Overfitting
Overfitting occurs when AI models fit too closely to training data, reducing prediction performance on new data, and is a critical issue that must be carefully managed when building quality prediction and defect detection systems in manufacturing.
Feature Engineering
Feature engineering is the preprocessing step that transforms raw manufacturing data into meaningful input variables for AI models, where field knowledge-based feature design is key to predictive model performance.
Explainable AI (XAI)
Explainable AI (XAI) is technology that transparently explains AI's decision-making processes, providing rationale for quality inspection, predictive maintenance, and process optimization in manufacturing, building operator trust and verifying safety.