Rancangan 100 Judul pembahasan ringan dan modular dalam rangkaian roadmap pembelajaran mengenai implementasi Explainable Artificial Intelligence pada Agrifood Supplychain. Ikuti pembahasannya di blog ini, dan chanel youtube : @kelasrahmathidayat https://www.youtube.com/@kelasrahmathidayat

  Part 1: Understanding XAI (Fundamentals)

1. What Is Explainable AI? A Gentle Introduction for Agrifood Professionals

2. Why Black-Box AI Is Risky for Food Supply Chain Decisions

3. The Three Pillars of XAI: Transparency, Interpretability, and Explainability

4. Interpretability vs. Explainability: Key Differences Every Researcher Should Know

5. Who Needs XAI in the Agrifood Sector? Stakeholders and Use Cases

6. The Cost of Unexplained AI: Trust, Compliance, and Adoption Barriers

7. Intrinsic vs. Post-Hoc Explainability: Choosing the Right Approach

8. Global vs. Local Explanations: When to Use What in Food Systems

9. Model-Specific vs. Model-Agnostic XAI Methods: A Practical Overview

10. Evaluating Explanations: Fidelity, Stability, and Comprehensibility

Part 2: XAI Taxonomy (Classification Frameworks)

11. A Systematic Taxonomy of XAI Methods for Agrifood Applications

12. Visual Explanations: Saliency Maps, Feature Visualization, and Beyond

13. Textual & Rule-Based Explanations: From Decision Trees to Natural Language

15. Counterfactual Explanations: What If the Crop Yield Had Been Different?

14. Example-Based Explanations: Prototypes, Criticisms, and Influential Instances

16. Feature Attribution Methods: Ranking What Matters Most in Your Model

17. Surrogate Models: Explaining Complex Predictions with Simple Interpretable Models

18. Attention Mechanisms as Explanations: Opportunities and Pitfalls

19. Hybrid XAI Taxonomies: Combining Local and Global Methods for Supply Chains

20. Task-Based Taxonomy: Explanations for Classification, Regression, and Forecasting

Part 3: XAI in Various Fields (Cross-Domain Insights)

21. XAI in Healthcare: Lessons for Agrifood Safety and Traceability

22. What Finance Can Teach Us About Explainable Credit Scoring for Farmers

23. XAI for Autonomous Vehicles: Parallels with Agricultural Robotics

24. Explainable NLP for Food Audits and Compliance Documentation

25. XAI in Manufacturing: Transferable Concepts for Food Processing Lines

26. Energy Sector XAI: Similarities in Time Series Load Forecasting for Cold Storage

27. Retail & Consumer Analytics: Explaining Demand Shocks for Perishables

28. Climate Science XAI: Interpretable Models for Weather Impact on Crops

29. XAI in E-Commerce Recommendations: Personalizing Sustainable Food Choices

30. Cross-Domain Bridges: What Agrifood Can Adapt from Legal XAI Standards

Part 4: XAI in Agriculture (Crop & Farm Level)

31. Explaining Yield Predictions: Which Soil and Weather Features Drove the Forecast?

32. Interpretable Pest and Disease Detection from Leaf Images

33. XAI for Precision Irrigation: Why Did the Model Recommend Less Water?

34. Soil Health Monitoring: Explaining Nutrient Deficiency Classifications

35. Weed Detection with Explainable CNNs: Distinguishing Crop from Weed

36. Livestock Welfare Prediction: Making Sense of Sensor Data with SHAP

37. Fertilizer Recommendation Systems: Transparent Nutrient Management

38. Crop Type Mapping from Satellite Imagery with GradCAM Explanations

39. Explainable Drone-Based Spraying: When and Why to Target Specific Zones

40. Farm-Level Risk Assessment: Explaining Drought Vulnerability Predictions

Part 5: XAI in the Agrifood Supply Chain (Post-Farm to Fork)

41. Post-Harvest Loss Prediction: Explaining Spoilage Risks During Transport

42. Cold Chain Integrity: Interpretable Anomaly Detection in Temperature Logs

43. Warehouse Inventory Optimization: Why Did the Model Recommend Restocking?

44. Explainable Logistics Route Planning for Fresh Produce Delivery

45. Food Fraud Detection: Local Explanations for Supplier Risk Flags

46. Traceability Blockchain + XAI: Auditable, Explainable Food Provenance

47. Demand Forecasting for Retailers: Explaining Sudden Peaks in Perishable Sales

48. Price Transmission Along the Chain: Explaining Farm-to-Fork Price Dynamics

49. Food Safety Recall Prediction: Which Factors Most Influence Contamination Risk?

50. Sustainable Sourcing Verification: XAI for Ethical Supply Chain Auditing

Part 6: XAI for Predictive Analytics (General)

51. Opening the Black Box of Predictive Models in Agrifood Systems

52. Feature Importance for Predictive Maintenance of Milling Equipment

53. Explaining Regression vs. Classification Predictions in Post-Harvest Quality

54. Counterfactual Scenarios for What-If Analysis: Yield Under Alternative Practices

55. Robustness Checks: How Reliable Are Your Local Explanations?

56. Predictive Model Debugging Using XAI: Identifying Leaky Features

57. Ensemble Model Explanations: Aggregating SHAP Values Across Trees

58. XAI for Anomaly Prediction in Grain Storage Silo Sensors

59. Predictive Lead Time: Explaining Why the System Forecasts a Processing Delay

60. From Prediction to Action: Closing the Loop with Human-Centric Explanations

Part 7: XAI in Time Series Forecasting (Methodological Focus)

61. Why Time Series Forecasting Needs Different XAI Techniques

62. Temporal Feature Attribution: LIME for Multivariate Lagged Inputs

63. Explaining Recurrent Neural Network Forecasts with Time-Step Relevance

64. Seasonal Pattern Explanations: Decomposing Forecasts into Interpretable Components

65. Event Impact Analysis: Explaining How Holidays or Shocks Change Forecasts

66. Contextual Importance for Time Series: When Past Values Actually Matter

67. Evaluating Explanation Stability Across Rolling Forecast Windows

68. XAI for Probabilistic Forecasts: Understanding Prediction Intervals

69. Comparing Attribution Methods for LSTMs on Temporal Agrifood Data

70. Visualizing Attention Over Time: What Was the Model Focusing On Last Week?

Part 8: XAI in Rice Price Forecasting (Domain-Specific)

71. Rice Price Forecasting 101: Why Explainability Is Non-Negotiable for Policy

72. Key Drivers of Wholesale Rice Prices: A SHAP-Based Global Analysis

73. Local Explanations for Extreme Price Spikes: What Went Wrong This Month?

74. Seasonal Decomposition with XAI: Monsoon Effects on Rice Price Forecasts

75. Explaining Lag Effects: How Past Prices Influence Current Forecasts

76. External Shock Analysis: Using Counterfactuals to Explain COVID-19 Impacts

77. Feature Selection for Rice Models: When to Include Input Prices (Fertilizer, Fuel)

78. Comparing LIME and SHAP for Rice Price Time Series Explanations

79. Attention Maps for Rice Markets: Which Previous Weeks Were Most Predictive?

80. Communicating Rice Price Forecasts to Policymakers: Visual Explanations That Work

Part 9: XAI in Indonesian Rice Price Forecasting (Geographically Focused)

81. Why Indonesia’s Rice Markets Demand Locally Adapted XAI

82. Explaining Regional Price Disparities: Java vs. Eastern Indonesia

83. The Role of Bulog Interventions: Detecting Government Stock Impacts with SHAP

84. Local Explanations for Harvest Period Price Drops Across Indonesian Regions

85. Weather-Driven Explanations: La Niña Effects on Indonesian Rice Forecasts

86. Import Dependency Attribution: Explaining How Global Prices Enter Local Models

87. Festival & Ramadan Effects: Temporally Localized Explanations for Price Spikes

88. Transportation Cost Visibility: Explaining Inter-Island Price Variation

89. Policy Simulation with Counterfactuals: What If Fuel Subsidies Changed?

90. Building Trust with BULOG: Deploying Interpretable LSTMs for National Stock Decisions

Part 10: XAI for Image Analysis (Computer Vision)

91. GradCAM for Crop Disease Detection: Which Leaf Spots Matter Most?

92. LRP (Layer-Wise Relevance Propagation) for Fruit Grading Images

93. Segmenting Weeds from Crops: Explainable Semantic Segmentation Maps

94. DeepLIFT for Ripeness Assessment: Color and Texture Attribution

95. Counterfactual Image Explanations: What Would Make This Defect Disappear?

96. Saliency Maps for Drone-Based Pest Counting: Avoiding Spurious Correlations

97. Evaluating Fidelity of Vision Explanations in High-Stakes Sorting Lines

98. Attention Rollout for ViTs (Vision Transformers) in Plant Phenotyping

99. XAI for Hyperspectral Imaging: Explaining Nutrient Deficiency Predictions

100. From Pixels to Decisions: GradCAM vs. LRP on Post-Harvest Quality Control

Bonus Section: Dedicated Method Tutorials (Supplementary)

SHAP

- SHAP Fundamentals: Shapley Values for Agrifood Feature Attribution

- Visualizing SHAP: Waterfall, Force, and Summary Plots for Food Data

LIME

- LIME for Tabular Data: Explaining One Farm's Yield Prediction

- LIME on Time Series: Interpreting a Rice Price Forecast Locally

GradCAM & Vision

- GradCAM Illustrated: Heatmapping Disease Lesions in Cassava Leaves

- GradCAM for Supply Chain: Explaining Defect Detection in Sorting Machines

DeepLIFT & LRP

- DeepLIFT vs. SHAP: Comparative Attribution for Neural Networks

- LRP Rule Selection: Epsilon, Gamma, and Z Rules for Agrifood CNNs

Stable & Explainable Attention

- Stable Attention Mechanisms: Why Consistency Matters for Food Safety

- Explainable Attention for LSTMs: Interpreting Focus on Seasonal Events

Interpretable Multi-Variable LSTM (IMV LSTM)

- IMV LSTM Architecture: Built-In Interpretability for Multivariate Forecasting

- Case Study: IMV LSTM for Indonesian Rice Price Forecasting

- IMV LSTM vs. Post-Hoc LIME: Comparing Explanation Quality in Food Supply Chains

---