Q5 econ3333
Study these flashcardsFree · spaced repetition · iPhone, Android, Mac & web
All cards
1
what does this picutre show?
Deep neural network computer vision
2
The above image data set contains an image from zero to nine, and a deep neural network classifies its label condition based on pixel information.
In this case, it is an example of _____________________.
binary classification
regression
multi-class classification
In this case, it is an example of _____________________.
binary classification
regression
multi-class classification
multiclass classification
3
The above figure is a structure of _____________________ and it is widely used for image classification. Additionally, it can be used for text classification.
Convolutional neural network
Deep neural network
Recurrent neural network
Artificial neural network
Convolutional neural network
Deep neural network
Recurrent neural network
Artificial neural network
Convolutional neural network
4
A________________ is a way of mapping text data into numerical vectors.
Backpropagation
Word embedding
Classification
Regression
Backpropagation
Word embedding
Classification
Regression
word embedding
5
The deep neural network (DNN) can classify the image dataset.
However, Image data is needed to be converted to a 2-dimensional matrix (28 rows x 28 columns = 768 pixels).
Then, it is needed to be converted into a vector for 768-pixels.
If the first fully connected layer (i.e., affine layer) has 100 hidden nodes, there will be 76,800 connections.
Therefore, the DNN model require high computation resources and time 1.______(True or False)
Therefore, the DNN model can learn spatial information in the image data 2._____ (True or False)
However, Image data is needed to be converted to a 2-dimensional matrix (28 rows x 28 columns = 768 pixels).
Then, it is needed to be converted into a vector for 768-pixels.
If the first fully connected layer (i.e., affine layer) has 100 hidden nodes, there will be 76,800 connections.
Therefore, the DNN model require high computation resources and time 1.______(True or False)
Therefore, the DNN model can learn spatial information in the image data 2._____ (True or False)
True, false
6
what type of image classification does the above figure represent?
Deep neural network (DNN)
7
what type of image classification does the above figure represent?
Convolutional Neural Network (CNN)
8
CNN models contains a convolutional and pooling layers. Convolutional layer have (1)____________ for a convolutional operation,
filter (i.e., a kernal)
9
(2) ____________ in CNN determines the number of pixels by which we move the filter across the input matrix as follows:
stride
10
(3) ____________ in CNN is adding a boundary around the input matrix to keep it the same size as follows:
padding
11
what is the above figure?
word2vec (W2V)
12
The above figure is the structure of the word2vec model in Skip-gram framework.
The word2vec predicts the neighbor words condition on a center words in a given window during the training
Which are true for the word2vec model?
W2V will find hidden topics from text data.
W2V learns its embedding vectors based on neighbor words in a given window
The dimension of W2V embedding vector is pre-determined as a hyperparameter
W2V will generate similar word embedding vectors for similar words (e.g., Coke and Pepsi)
Word2vector (W2V) is a word distribution-based embedding method
The word2vec predicts the neighbor words condition on a center words in a given window during the training
Which are true for the word2vec model?
W2V will find hidden topics from text data.
W2V learns its embedding vectors based on neighbor words in a given window
The dimension of W2V embedding vector is pre-determined as a hyperparameter
W2V will generate similar word embedding vectors for similar words (e.g., Coke and Pepsi)
Word2vector (W2V) is a word distribution-based embedding method
W2V learns its embedding vectors based on neighbor words in a given window
The dimension of W2V embedding vector is pre-determined as a hyperparameter
W2V will generate similar word embedding vectors for similar words (e.g., Coke and Pepsi)
Word2vector (W2V) is a word distribution-based embedding method
The dimension of W2V embedding vector is pre-determined as a hyperparameter
W2V will generate similar word embedding vectors for similar words (e.g., Coke and Pepsi)
Word2vector (W2V) is a word distribution-based embedding method
13
In this regards, each word can have multiple different w2v embedding vectors 1.____________(True, False)
In addition, similar words will have (2)_________(similar, different) W2V wording embedding for each other.
Lastly, similar words will be positioned in (3)___________(similar, different) locations in the above 2 dimensional real-vector space (i.e., Euclidian space)?
In addition, similar words will have (2)_________(similar, different) W2V wording embedding for each other.
Lastly, similar words will be positioned in (3)___________(similar, different) locations in the above 2 dimensional real-vector space (i.e., Euclidian space)?
false, similar, similar
14
What is not true for Word2vec model?
Word2vector (W2V) is a word distribution-based embedding method.
The dimension of the W2V embedding vector is determined by the number of entire words.
W2V generates a dense real vector embedding for each word.
W2V will generate similar word embedding vectors for similar words (e.g., Coke and Pepsi)
Word2vector (W2V) is a word distribution-based embedding method.
The dimension of the W2V embedding vector is determined by the number of entire words.
W2V generates a dense real vector embedding for each word.
W2V will generate similar word embedding vectors for similar words (e.g., Coke and Pepsi)
The dimension of the W2V embedding vector is determined by the number of entire words.
15
What is not a key hyperparameter for a convolutional neural network (CNN) model?
The number of filters
Filter sizes
The dimension of word embedding vector
The number of epochs
Learning rate
The number of filters
Filter sizes
The dimension of word embedding vector
The number of epochs
Learning rate
The dimension of word embedding vector
16
The Word2Vec model has two different types: the Continuous Bag of Words (CBOW) or Skip-gram.
CBOW predicts the center word condition on neighbor words in a given window during the training.
Skip-gram predicts the neighbor words condition on a center words in a given window during the training.
The above case in figure is _____________________________
CBOW word2vec
Skip-gram word2vec
CBOW predicts the center word condition on neighbor words in a given window during the training.
Skip-gram predicts the neighbor words condition on a center words in a given window during the training.
The above case in figure is _____________________________
CBOW word2vec
Skip-gram word2vec
CBOW word2vec
17
The above figure is the case of ______________________________________________ from text data.
The CNN on the top W2V embedding for text classification.
The W2V model on the top of CNN embedding for text classification.
The W2V model on the top of CNN embedding for regression
The CNN on the top W2V embedding for regression
The CNN on the top W2V embedding for text classification.
The W2V model on the top of CNN embedding for text classification.
The W2V model on the top of CNN embedding for regression
The CNN on the top W2V embedding for regression
The CNN on the top W2V embedding for text classification.
18
We can add other embedding vectors from structured data and unstructured data (image, voice) on the last layer of the CNN.
This type of approach is called ____________(more than one type of data is used).
an autoencoder-modal approach
a regression-modal approach
a classification-modal approach
a multi-modal approach
This type of approach is called ____________(more than one type of data is used).
an autoencoder-modal approach
a regression-modal approach
a classification-modal approach
a multi-modal approach
a multi-modal approach