Explore Your Understanding

Match the Concepts

If you understood this section well, you should be able to match the concepts effectively. This exercise will help reinforce your learning. Try to match each technique/Concept with its corresponding description before clicking the 'Answer' button. Once you’ve made your matches, click 'Answer' to see how well you did!

Duration: 15-20 minutes

1. Techniques & core definitions

Technique	Description
A. Transfer Learning	1. Uses task-specific knowledge to predict new, unseen categories
B. Active Learning	2. Selects the most informative unlabeled samples for annotation
C. Zero-Shot Learning	3. Integrates prior biophysical or ecological models
D. Process-Aware Learning	4. Uses pretrained model weights as initialization

Answer

Answer Key:
• A – 4
• B – 2
• C – 1
• D – 3

2. Learning approaches & label strategies

Technique	Description
A. Weakly Supervised Learning	1. Combines a small amount of labeled data with a large amount of unlabeled data
B. Semi-Supervised Learning	2. Uses coarse, noisy, or incomplete labels to train models
C. Self-Supervised Learning	3. Learns representations using pseudo-labels or pretext tasks
D. Few-Shot Learning	4. Learns to generalize from very few examples per class

Answer

Answer Key:
A – 2
B – 1
C – 3
D – 4

3. Application contexts

Technique	Example use case
A. Active Learning	1. A system flags uncertain areas in an image for expert labeling
B. Ensemble Learning	2. Multiple models trained on different seasons are combined
C. Process-Aware Learning	3. A model learns using crop growth simulations alongside satellite data
D. Zero-Shot Learning	4. A model identifies a new crop class it wasn’t trained on

Answer

Answer Key:
A – 1
B – 2
C – 3
D – 4

4. What problem does it solve?

Technique	Solves which challenge?
A. Few-Shot Learning	1. Learning with limited samples for rare crop types
B. Transfer Learning	2. Adapting models trained in one region to another
C. Active Learning	3. Reducing annotation costs by prioritizing sample selection
D. Ensemble Learning	4. Improving model robustness and generalization

Answer

Answer Key:
A – 1
B – 2
C – 3
D – 4

5. Key components in evaluation

Concept	Description
A. Spatial k-Fold CV	1. Ensures validation data are geographically independent from training
B. Random Sampling	2. Often results in spatial data leakage
C. Cross-Validation Fold	3. A subset used for model evaluation
D. Model Generalisation	4. Ability to apply a model to new, unseen regions

Answer

Answer Key:
A – 1
B – 2
C – 3
D – 4

Short-Answer Critical Thinking

Critical thinking is essential for deep understanding. In this section, you will encounter descriptive questions that challenge your comprehension of the material. Take a moment to formulate your answers before revealing the correct responses. Click 'Answer' to check your understanding and see how many you got right!

Duration: 10-15 minutes

1. Why might self-supervised learning be particularly beneficial for agricultural imagery?

Answer

Expected answer: Because large quantities of unlabeled data are available, and self-supervised methods can extract useful representations without manual annotation.

2. Explain how process-aware learning differs from other deep learning strategies discussed.

Answer

Expected answer: It incorporates domain knowledge from physical or ecological models to guide learning, rather than relying solely on data-driven learning.

3. What risks might arise if spatial autocorrelation is not accounted for during model evaluation?

Answer

Expected answer: Models may appear to perform well on validation data that is too similar to the training data, leading to over-optimistic results that don’t generalize spatially.