2.1. Instance Based Leraning Flashcards
What is Instance Based Learning (IBL)?
A type of learning where the model makes predictions based on specific instances from the training data.
How does IBL differ from traditional learning methods?
IBL does not create a general model; it relies on specific examples for predictions.
What is the role of distance metrics in IBL?
They determine how similarity between instances is measured (e.g., Euclidean distance).
What is the k-Nearest Neighbors (k-NN) algorithm?
An IBL method that classifies instances based on the majority class of their k nearest neighbors.
What are the advantages of Instance Based Learning?
Simplicity, adaptability to new data, and no need for a training phase
What are the disadvantages of IBL?
High memory usage and computational cost during prediction.
How does IBL handle noisy data?
It can be sensitive to noise, as it relies heavily on specific instances.
What is the concept of “lazy learning” in IBL?
The model does not generalize until a query is made, delaying computation until necessary.
How can IBL be improved?
By using techniques like instance weighting or pruning irrelevant instances.
What is the significance of the training set size in IBL?
A larger training set can improve the accuracy of predictions but increases computational cost
What is the impact of dimensionality on IBL?
High dimensionality can lead to the “curse of dimensionality,” making distance calculations less meaningful.
How does IBL perform in multi-class classification?
It can effectively handle multi-class problems by considering the nearest neighbors across all classes.
What is the role of feature selection in IBL?
Selecting relevant features can enhance the performance and efficiency of the model.
How does IBL relate to clustering?
Both methods can group similar instances, but IBL focuses on classification based on specific examples.
What applications are suitable for Instance Based Learning?
Image recognition, recommendation systems, and any domain where specific instance similarity is crucial.
