QCM : Fundamentals of Image Classification and Neural Networks — 9 questions

Question 1

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

Vladimir Vapnik in 1995

Yann LeCun in 1989

Yoshua Bengio in 2013

Geoffrey Hinton in 2006

Explication

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.

Answer

Vladimir Vapnik in 1995

Question 2

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

To derive the final class label based on confidence levels

To generate feature descriptors for images

To implement data augmentation techniques

To pre-process images before classification

Explication

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.

Answer

To derive the final class label based on confidence levels

Question 3

3. What is image classification primarily about?

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

Segmenting an image into different regions based on texture

Enhancing image resolution for better visual quality

Detecting the presence of objects within an image

Explication

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.

Answer

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

Question 4

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

SIFT + Bag of Visual Words (BoVW)

HOGs or LBPs

SURF + BoVW

Interest point detection only

Explication

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

Answer

HOGs or LBPs

Question 5

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

Vladimir Vapnik in 1995

Geoffrey Hinton in 2006

Yann LeCun in 1989

Alex Krizhevsky in 2012

Explication

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

Answer

Vladimir Vapnik in 1995

Question 6

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

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

It analyzes entire images globally to define features

It uses deep neural networks to automatically learn features

It applies wavelet transforms to extract features from images

Explication

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

Answer

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

Question 7

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

Linear classifiers

Neural networks

K-Nearest Neighbors (KNN)

Clustering algorithms like K-Means

Explication

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

Answer

Clustering algorithms like K-Means

Question 8

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

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

It initializes the neural network weights randomly

It performs forward pass calculations only

It increases the learning rate to improve training

Explication

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

Answer

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

Question 9

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

ReLU (Rectified Linear Unit)

Linear activation

Softmax only at the output layer

Identity function

Explication

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

Answer

ReLU (Rectified Linear Unit)

QCM : Fundamentals of Image Classification and Neural Networks — 9 questions

Questions et réponses du QCM

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

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

3. What is image classification primarily about?

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

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

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

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

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

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

Révisez avec les flashcards

Approfondir avec la fiche

Cours similaires

Introduction à l'Intelligence Artificielle Symbolique

Introduction aux documents et sécurité en construction

Introduction à l'Informatique et IA

Introduction aux marchés publics et réglementations du bâtiment

Gestion des imprimantes Windows et PDF

Introduction aux bases de données relationnelles

Crée tes propres QCM