Cuestionario: Fundamentals of Image Classification and Neural Networks — 9 preguntas

1. Who introduced the Support Vector Machine (SVM) model and in which year?

A. Vladimir Vapnik in 1995

B. Yann LeCun in 1989

C. Yoshua Bengio in 2013

D. Geoffrey Hinton in 2006

Answer: (A)

Explicación

Vladimir Vapnik is credited with introducing Support Vector Machines in 1995, as explicitly mentioned in the content. The other options list prominent researchers in machine learning but are associated with different contributions or dates.

2. What is the primary purpose of class scores estimation in image classification?

A. To derive the final class label based on confidence levels

B. To generate feature descriptors for images

C. To implement data augmentation techniques

D. To pre-process images before classification

Answer: (A)

Explicación

Class scores estimation predicts numerical confidence levels for each class, which are then used to assign the most probable class label, not to generate features or process images.

3. What is image classification primarily about?

A. Assigning a class label to an object in an image based on features and class scores

B. Segmenting an image into different regions based on texture

C. Enhancing image resolution for better visual quality

D. Detecting the presence of objects within an image

Answer: (A)

Explicación

Image classification is primarily about identifying the class of an object within an image, often by estimating class scores based on features, and then assigning the most likely label.

4. Which feature extraction method is an example of using global features in image classification?

A. SIFT + Bag of Visual Words (BoVW)

B. HOGs or LBPs

C. SURF + BoVW

D. Interest point detection only

Answer: (B)

Explicación

HOGs, LBPs, and Haar wavelets are examples of global features that describe the entire image, unlike local features like SIFT or SURF.

5. Who introduced the Support Vector Machine (SVM) model and in which year?

A. Vladimir Vapnik in 1995

B. Geoffrey Hinton in 2006

C. Yann LeCun in 1989

D. Alex Krizhevsky in 2012

Answer: (A)

Explicación

Vladimir Vapnik introduced the SVM model in 1995, which became a foundational method in machine learning for classification tasks.

6. What is the main idea behind the Bag of Visual Words (BoVW) model?

A. It extracts local features, clusters them into codewords, and represents images as histograms of these codewords

B. It analyzes entire images globally to define features

C. It uses deep neural networks to automatically learn features

D. It applies wavelet transforms to extract features from images

Answer: (A)

Explicación

BoVW involves extracting local features, clustering them, and representing images with histograms of these codewords, capturing local patterns for classification.

7. Which of the following is NOT a commonly used classifier for image features?

A. Linear classifiers

B. Neural networks

C. K-Nearest Neighbors (KNN)

D. Clustering algorithms like K-Means

Answer: (D)

Explicación

Clustering algorithms like K-Means are unsupervised and not typically used as classifiers directly for labeled image features, unlike linear classifiers, NNs, or KNN.

8. In the context of neural network training, what role does backpropagation play?

A. It updates the network weights by propagating errors backward through the network

B. It initializes the neural network weights randomly

C. It performs forward pass calculations only

D. It increases the learning rate to improve training

Answer: (A)

Explicación

Backpropagation computes gradients of the loss function with respect to each weight, enabling the network to learn by updating weights to minimize errors.

9. Which activation function is frequently used in neural networks to introduce non-linearity?

A. ReLU (Rectified Linear Unit)

B. Linear activation

C. Softmax only at the output layer

D. Identity function

Answer: (A)

Explicación

ReLU is a popular activation function because it introduces non-linearity and helps neural networks learn complex patterns efficiently.