A4.1 Machine Learning

1234567891011121314151617181920212223242526272829
Across
  1. 2. Describes data where each example has the correct output attached
  2. 4. Medical ___ diagnostics: using ML to spot disease in scans
  3. 5. Type of machine learning that finds patterns in data with no correct answers provided
  4. 7. ___ analysis: deciding whether a review is positive or negative
  5. 10. Chip whose logic can be reconfigured after manufacture
  6. 15. Type of learning that improves through trial and error using rewards and penalties
  7. 17. Assigning an input to a category, such as naming the object in a photo
  8. 19. ___ network: connected nodes loosely modelled on the brain
  9. 21. A deep network has many hidden ones between input and output
  10. 22. Abbreviation for centres of clustered supercomputers used for the largest jobs
  11. 25. Platform type offering on-demand remote computing over the internet
  12. 26. Type of processing where many calculations run at the same time
  13. 27. Object ___: locating items within an image, as in self-driving cars
  14. 28. Compute-heavy phase in which a model learns from data
  15. 29. ___ learning: uses networks with many hidden layers to learn complex features
Down
  1. 1. Feedback signal that a trial-and-error learner tries to maximise
  2. 3. Market ___ analysis: finding products that customers often buy together
  3. 6. ___ device: runs a model locally, close to where the data is collected
  4. 7. Hardware requirement for holding large datasets and saved models
  5. 8. Type of machine learning that learns from input-output pairs with known correct answers
  6. 9. Chip custom-built for one task: very efficient but cannot be reprogrammed
  7. 11. Delay between input and response, kept low by processing locally
  8. 12. ___ learning: reuses a model built for one task as the starting point for a related task
  9. 13. Ability of hardware to cope with growing data or demand
  10. 14. Phase in which a finished model makes predictions on new data
  11. 15. Field where machines learn to navigate physical spaces
  12. 16. Processor with thousands of cores, widely used to train models
  13. 18. The decision-maker that takes actions in an environment and learns from feedback
  14. 20. Describes a model that has already learned from a large dataset before being fine-tuned
  15. 23. Chip designed by Google specifically for tensor operations
  16. 24. Abbreviation for the application area that handles human text and speech