Dr. Tiziana Ligorio x CSCI 493.77 - Deep Learning, Hunter College of the City University of New York
Model evaluation for regression
Model evaluation for classification
The model receives a set of inputs along with the corresponding correct outputs, and the algorithm learns by comparing its actual output with correct outputs to find errors. It then modifies the model accordingly (model training).
<aside> ⚠️
This image illustrates a simplistic (traditional) approach to ML. In reality the cycle is often more complex, involving reiterating through problem definition and data processing as well. For now, this simple diagram will be enough to introduce the topic.
</aside>
Learning rate: how fast the model adapts to new data.
High learning rate: rapidly adapt to new data but also forget old data
Low learning rate: learns slowly but less sensitive to noise in data and outliers
$$ \epsilon $$
The system cannot learn a predictive model if dataset is too small, if data is not representative, noisy or polluted with irrelevant features