Thuật ngữ AI
Từ điển đầy đủ về Trí tuệ nhân tạo
Global interpretation methods
Techniques that explain the overall and general behavior of a model across the entire dataset
Local interpretation methods
Approaches that explain individual predictions for specific observations
Feature Attribution
Techniques quantifying the importance and impact of each feature on model predictions
Interpretability of deep models
Specialized methods for understanding and explaining deep neural networks
Interpretability Visualization
Graphical and visual representations to facilitate understanding of model decisions
Specific methods for models
Interpretation techniques adapted to particular types of models such as decision trees
Counterfactual interpretability
Generation of counterexamples to explain why a specific prediction was made
Interpretability Evaluation Metrics
Indicators and measures to quantify the quality and reliability of model explanations
Causal interpretability
Analysis of cause-effect relationships in AI model decisions
Interpretability for ensemble models
Specific techniques for interpreting random forests, gradient boosting and other ensemble models
Temporal interpretability
Methods for explaining models working on sequential and temporal data
Multimodal interpretability
Approaches to explain models combining multiple data types (text, image, sound)
Contrastive interpretability
Techniques comparing model decisions across different scenarios and conditions
Explanations by prototypes
Methods based on identifying representative examples to explain predictions
Context-sensitive interpretability
Approaches adapting explanations to the application domain and target audience