Explane Transfer-Learning
traditional ML: no knowlegde transfer
What is the Sim-to-Real gap?
Name advantages and disadvantages of simulated data
Simulation-assisted machine learning
machine-learning assisted simulation
Das mit den dreiecken
Physics-informed Machine Learning
Semi-supervised Learning?
What assumptions are made?
Small amout of labeled data available but a lot of unlabeled data (e.g. medical images)
Active Learning
Choose unlabeled data point, ask for help
Advantage: Use more and cheaper unlabeled data
Difference between Incremental learning and online learning?
Online learning processes data in real time and continuously updates its model, while incremental learning processes chunks of data at scheduled intervals.
Examples:
online: Stockprices online
incremental: Picture recognition (airplane, car, bird … tree)
Order the following machine learning models by degree of interpretability.
Deep NN,
SVM,
K-NN,
Lin. Reg
Decision tree
Key Factors for AI Reliability and Resilience:
High data quality
Robust algorithms
Testing and validation
Human oversight and intervention
AI technology for agriculture
Agriculture functions
Sustainability groth objektives
Major fields where AI- and data-driven Methods have a sustainable impact
Agriculture (e.g., crop prediction, plant monitoring, autonomous vehicles, weather forecasting)
Energy (e.g., building energy management, lightning systems, renewable energy)
Transport (e.g., traffic control, good delivery, prediction of traffic pile up)
Manufacturing (e.g., predictive maintenance, quality control, demand planning)
diagramm: How do real world and simulation interact?
Zuletzt geändertvor 4 Monaten
