VIP20
本课程属于：005-Artificial Intelligence and Machine Learning Engineering English Technical Courses · 共 20 门相关课程，组合学习更完整。
查看所属VIP全部课程 →
← 返回所属目录 咨询 / 购买

Explainable AI for High-Stakes Engineering Decisions

30 Chapters
View Course Samples Online
01
High-stakes engineering decisions
defining risk, accountability, and the role of AI in safety-critical, regulated, and high-consequence domains.
High-stakesdecisions
02
Explainability requirements from engineering standards and regulations
ISO 26262, IEC 61508, FDA SaMD, EU AI Act, and NIST AI RMF.
Explainabilityrequirements
03
Taxonomy of explainable AI methods
intrinsic vs. post-hoc, model-specific vs. model-agnostic, local vs. global, and their suitability for engineering decisions.
Taxonomyexplainable
04
Feature attribution techniques for engineering models
SHAP, LIME, Integrated Gradients, and their interpretation in physical units and design variables.
attributiontechniques
05
Surrogate models and rule extraction for high-stakes decisions
decision trees, rule lists, and symbolic regression as interpretable approximations.
Surrogateextraction
06
Counterfactual explanations for engineering design
generating actionable alternatives under physical and safety constraints.
Counterfactualexplanations
07
Concept-based explanations
mapping latent representations to engineering concepts such as stress, fatigue, and failure modes.
Concept-basedexplanations
08
Uncertainty quantification and calibration for trustworthy explanations
Bayesian methods, conformal prediction, and their role in decision confidence.
Uncertaintyquantification
09
Evaluating explanation quality
faithfulness, stability, robustness, and human-grounded metrics for engineering stakeholders.
Evaluatingexplanation
10
Human factors in explainable AI for engineers
cognitive biases, trust calibration, and decision-making under time pressure.
explainableengineers
11
Case study
explainable AI for structural health monitoring and failure prediction in civil infrastructure.
explainablestructural
12
Case study
explainable AI for autonomous vehicle perception and trajectory planning decisions.
explainableautonomous
13
Case study
explainable AI for medical device control and clinical decision support in high-risk procedures.
explainableclinical
14
Case study
explainable AI for power grid stability and fault diagnosis in energy systems.
explainablestability
15
Case study
explainable AI for aerospace flight control and anomaly detection in avionics.
explainableaerospace
16
Case study
explainable AI for industrial process control and chemical plant safety.
explainableindustrial
17
Explainability in model-based engineering
integrating XAI with digital twins, simulation, and physics-based models.
Explainabilitymodel-based
18
Explainability for reinforcement learning in safety-critical control
policy summarization, reward decomposition, and safe exploration.
Explainabilityreinforcement
19
Explainability for deep learning in engineering
saliency maps, attention, and layer-wise relevance propagation for sensor and image data.
Explainabilitysaliency
20
Explainability for ensemble and hybrid models
random forests, gradient boosting, and physics-informed neural networks.
Explainabilityensemble
21
Adversarial robustness of explanations
attacks, manipulation, and defensive strategies for high-stakes settings.
Adversarialrobustness
22
Fairness, bias, and ethical considerations in explainable AI for engineering decisions.
Fairness, bias, and ethical considerations in explainable AI for engineering decisions.
Fairnessconsiderations
23
Interactive and visual analytics for explainable AI
dashboards, what-if analysis, and human-in-the-loop decision support.
Interactiveanalytics
24
Explanation interfaces for operators and maintainers
designing for actionability, timeliness, and regulatory audit.
Explanationinterfaces
25
Verification and validation of explainable AI systems
testing explanation correctness, completeness, and consistency.
Verificationvalidation
26
Documentation and reporting for regulatory compliance
model cards, explanation reports, and audit trails.
Documentationreporting
27
Deployment challenges
latency, scalability, and integration of XAI into real-time engineering workflows.
Deploymentchallenges
28
Organizational and process integration
embedding XAI into design reviews, risk assessments, and incident investigations.
Organizationalintegration
29
Emerging trends
causal explainability, neuro-symbolic AI, and foundation models for engineering explanations.
Emergingexplainability
30
Capstone project
designing, implementing, and evaluating an explainable AI solution for a chosen high-stakes engineering decision scenario.
Capstonedesigning