Topic A4 is Higher Level (HL) only. This unit looks at where machine learning is used in the real world, at two unsupervised approaches (clustering and association rules), and at the ethical issues the course stresses, above all bias in training data and how to reduce harm.
Slide 1 / 4
Real-world applications
Supervised learning with deep learning, especially CNNs, powers medical imaging diagnostics and object detection and classification, where models learn from labelled images.
Natural language processing uses deep learning (transformers and RNNs) and transfer learning for translation, sentiment analysis and chatbots; models such as BERT are pre-trained then fine-tuned.
Reinforcement learning is central to robotics navigation, where a robot learns sequences of decisions through rewards and penalties.
Sentiment analysis classifies text as positive, negative or neutral, and market basket analysis finds which products are likely to be bought together.
Check you've got it
0/4
1. Which learning paradigm does the book say best fits robotics navigation?
2. What does k-means clustering use to represent the centre of each cluster?
3. In association rule learning, which measure indicates how reliable a rule such as X implies Y is?
4. Why is bias in the training data an ethical concern in machine learning?
Free revision content. Multiple-choice answers are marked instantly. Written answers are marked by AI on /free-mark.