Advanced Algorithm Flashcards
What is the name of the machine learning technique that allows a neural network to focus on specific parts of an input sequence?
a) Attention Model
b) KNN Model
c) Support vector machine Model
d) Random Forest Model
a) Attention Model
What are the two main steps of the Attention Mechanism?
a) Make a separate line between different classes
b) Calculating the attention weights and generating the context vector.
c) Focus in the First part of the input sequence and ignore the rest
d) None of the above
b) Calculating the attention weights and generating the context vector.
What is the advantage of using the Attention Mechanism over a traditional sequence-to-sequence model?
a) Calculating the Weights and Softmax.
b) Focus in the First part of the input sequence and ignore the rest.
c) The Attention Mechanism lets the model focus on specific parts of the input sequence.
d) Make it difficult in Training data
c) The Attention Mechanism lets the model focus on specific parts of the input sequence.
Advantages of Attention Mechanism
a) Reduced Information Loss
b) Improved Performance
c) Interpretability
d) All of the above
d) All of the above
……………………. captures the relevant information from the input sequence needed to generate the output at each time step, allowing the model to focus on different parts of the input sequence as needed.
a) Context Vector
b) Softmax function
c) associated energy
d) global alignment weights
a) Context Vector
What is a Restricted Boltzmann Machine (RBM)?
a) An unsupervised learning deep neural network type
b) A classification task supervised learning algorithm
c) A reinforcement learning model for decision-making
d) A specific kind of regression problem support vector machine
a) An unsupervised learning deep neural network type
Which of the following statements about RBMs is true?
a) RBMs are either fully connected neural networks with no connection
b) limitations or shallow neural networks with only one hidden layer.
c) RBMs are generative models capable of learning an input data’s probability distribution.
d) Supervised learning problems are the main applications of RBMs
c) RBMs are generative models capable of learning an input data’s probability distribution.
What is the key characteristic of the “restricted” nature of RBMs?
a) Connections between visible and hidden units are bidirectional
b) Connections within the visible and hidden layers are sparse
c) Connections between visible units are limited to nearest neighbors
d) Connections between visible and hidden units are not allowed within the same layer
d) Connections between visible and hidden units are not allowed within the same layer
RBMs are trained using which algorithm?
a) Backpropagation
b) Gradient descent
c) Contrastive divergence
d) K-means clustering
c) Contrastive divergence
Which task is RBM commonly used for?
a) Image classification
b) Speech recognition
c) Collaborative filtering
d) Natural language processing
c) Collaborative filtering
In an RBM, what is the purpose of the hidden layer?
a) To reconstruct the input data
b) To capture latent features in the data
c) To perform dimensionality reduction
d) To calculate the error between predicted and actual outputs
b) To capture latent features in the data
Which activation function is commonly used in the hidden layer of an RBM?
a) Sigmoid
b) Relu
c) Tanh
d) Linear
a) Sigmoid
Which of the following is NOT a potential application of RBMs?
a) Collaborative filtering for recommendation systems
b) Dimensionality reduction in feature space
c) Image classification using convolutional RBMs
d) Reinforcement learning for game playing
d) Reinforcement learning for game playing
What is diffusion model?
a) The process of particles moving from an area of low concentration to an area of high concentration.
b) The process of particles, information, or energy moving from an area of high concentration to an area of lower concentration.
c) The process of creating new data samples using a stochastic process.
d) The process of transforming noisy data into clean data samples.
b) The process of particles, information, or energy moving from an area of high concentration to an area of lower concentration.
What are diffusion models in machine learning?
a) Models that generate new data based on the data they are trained on.
b) Models used for image colorization and style transfer.
c) Models that simulate a diffusion process to transform noisy data into clean data
samples.
d) Models that estimate the likelihood of data samples using the score function.
a) Models that generate new data based on the data they are trained on.
Which type of diffusion model is used for probabilistic data generation?
a) Score-Based Generative Models (SGMs)
b) Stochastic Differential Equations (SDEs)
c) Denoising Diffusion Probabilistic Models (DDPMs)
d) Forward Diffusion Models
c) Denoising Diffusion Probabilistic Models (DDPMs)
What is the purpose of data preprocessing in diffusion models?
a) To generate high-quality images with realistic textures.
b) To handle missing data during the generation process.
c) To transform images from one style to another.
d) To prepare the data for subsequent transformations during the diffusion process.
d) To prepare the data for subsequent transformations during the diffusion process.
How do diffusion models generate new data samples?
a) By applying a sequence of invertible transformations to diffuse the data.
b) By estimating the score function of the data distribution.
c) By simulating a diffusion process that transforms noisy data into clean data samples.
d) By applying a sequence of reverse transformations to map the data back to a simple distribution.
a) By applying a sequence of invertible transformations to diffuse the data.
What does BERT stand for in Natural Language Processing (NLP)?
a) Bidirectional Encoder Representations from Transformers
b) Basic Encoding Representations for Text
c) Binary Embedding Representations for Training
a) Bidirectional Encoder Representations from Transformers
How does BERT achieve bidirectionality in understanding text?
a) It reads text from left to right only.
b) It uses Transformer models with attention mechanisms.
c) It relies on recurrent neural networks for context understanding.
b) It uses Transformer models with attention mechanisms.
What are some key advantages of BERT in NLP applications?
a) Handling long-range dependencies and context understanding.
b) Generating high-resolution images from textual descriptions.
c) Performing real-time sentiment analysis on social media data.
a) Handling long-range dependencies and context understanding.
Which pre-training tasks are commonly used to train BERT models?
a) Image classification and object detection.
b) Masked Language Model (MLM) and Next Sentence Prediction
c) Clustering and dimensionality reduction.
b) Masked Language Model (MLM) and Next Sentence Prediction
What are some popular variants or adaptations of BERT used in specific domains?
a) Bio BERT for biomedical text analysis.
b) Geo BERT for geographical information extraction.
c) Music BERT for music recommendation systems.
a) Bio BERT for biomedical text analysis.
What is the definition of GPT?
a) A natural language models.
b) A programming language.
c) A type of computer.
d) A search engine.
a) A natural language models.
