Quiz9
Deep Learning and Computer Vision Quiz
Test your knowledge on computer vision and neural networks with our engaging quiz! This quiz covers a variety of topics including algorithms, neural network architectures, and classifiers.
- 12 thought-provoking questions
- Multiple choice and checkbox formats
- Perfect for students and professionals alike
Which are true regarding the Viola-Jones face detector?
It only detects real human faces
It extracts Haar-like features from the images
It's very robust for side faces
Can be adapted to detect other things
I want to create my onw Cascade Classifier of British car license plates. What will I need to do?
Train the classifier with thousands of images of British license plates
Train the classifier with thousands of images of British license plates, and thousands of something else
Train the classifier with hundreds of images of British license plates
Train the classifier with thousands of images of British license plates, and thousands of Portuguese license plates
Regarding the cascade approach used in the Viola-Jones face detector, which statements are true?
The cascading process doesn't use any Haar-like features during classification
Only images "patches" that will highly likely have a face on it will be allowed to go through the cascade
There are Haar-like features than can tell me with a great certainty that an image "patch" doesn't contain a face; those are found during training and, then, used first in the cascade during classification
Usually, the majority of "patches" in one image won't contain a face, therefore, rejecting those immediately, will prevent calculating more detailed Haar-like features on them, resulting in great performance
What does a non-covolutional Artificial Neural Network needs to adjust during the learning (i.e. training) phase?
The training dataset
The contributions (i.e. weights) of each neuron in the previous layer to each neuron in the new layer
The labels
The bias of each neuron
What does a Convolutional Neural Network needs to adjust during the learning (i.e. training) phase?
The contributions (i.e. weights) of each neuron in the previous layer to each neuron in the new layer, when talking about dense layers
The contributions (i.e. weights) of each neuron in the previous layer to each neuron in the new layer, when talking about convolutional layers
The labels
The values of the filters, when talking about the convolutional layers
N the context of Neural Networks what is the idea behind the technique known as Dropout?
It consists in randomly "shutting down" some neurons during testing, so that the model doesn't overfit to the training data (i.e. Very clever at recognizing the training data, but very stupid at recognizing unseen data), with the benefit that it also results in better performance during training
Even without looking at any of the labels of the dataset, why can I immediately reach the conclusion that the model is not expecting one-hot encoded labels? model = tf.keras.models.Sequential([ tf.keras.layers.Flatten(input_shape=(28, 28)), tf.keras.layers.Dense(128, activation='relu'), tf.keras.layers.Dropout(0.2), tf.keras.layers.Dense(10, activation='softmax') ]) model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
The loss function is the Sparse Categorical Cross Entropy
The loss function is the Categorical Cross Entropy
The activation function of the output layer is the Softmax
It uses the Adam Optimizer
How can I tell that we are dealing with a multi-classification problem, just by looking a the follwing model definition? model = tf.keras.models.Sequential([ tf.keras.layers.Flatten(input_shape=(28, 28)), tf.keras.layers.Dense(128, activation='relu'), tf.keras.layers.Dropout(0.2), tf.keras.layers.Dense(10, activation='softmax') ]) model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])
It uses the Adam Optimizer
The loss function is the Sparse Categorical Cross Entropy
The activation function of the output layer is the Softmax
The activation function of the output layer is the Sigmoid
What is purpose of the definition of the first "layer" in the following model? model = tf.keras.models.Sequential([ tf.keras.layers.Flatten(input_shape=(28, 28)), tf.keras.layers.Dense(128, activation='relu'), tf.keras.layers.Dropout(0.2), tf.keras.layers.Dense(10, activation='softmax') ])
Convert a 28×28 color image into a vector of 2352 elements
Convert a 28×28 grayscale image into a 3D vector of 784 elements
Convert a 28×28 grayscale image into a vector of 784 elements
Tests performed on a given classifier achieved an AUC (Area Under the Curve) of 0.96. The result is very promising. True or false?
True
False
{"name":"Quiz9", "url":"https://www.quiz-maker.com/QPREVIEW","txt":"Test your knowledge on computer vision and neural networks with our engaging quiz! This quiz covers a variety of topics including algorithms, neural network architectures, and classifiers.12 thought-provoking questionsMultiple choice and checkbox formatsPerfect for students and professionals alike","img":"https:/images/course4.png"}
More Quizzes
AI/ML
2010103
Deep Learning
20100
AI Training Exam | WE
2412123
NeuroData Training : Deep Learning
10532
Artificial intelligence and machine learning - take the quiz
201052
Artificial intelligence
1587
Decision Trees Mastery Quiz
11616
Selected topics in computers
683453
Ultimate AI and Machine Learning Quiz
3417147
Machine Learning
840
AI Epi
2010113
Intro to Deep Learning : LR QUIZ
1050