Karteikarten: Fundamentals of Image Classification and Neural Networks — 9 Karten

Support vector machines — role?

Find optimal hyperplane with maximum margin.

Local features — purpose?

Capture details from image regions.

Neural network — basic structure?

Layers of neurons with weights, biases, activation functions.

Bag of visual words — process?

Clusters local features into codewords.

Neural network — training?

Adjust weights via backpropagation.

Activation functions — role?

Introduce non-linearity into neural networks.

Transfer learning — definition?

Using pre-trained models for new tasks.

Loss functions — purpose?

Quantify prediction error.

Class scores estimation — meaning?

Predict confidence for each class.