26 Questions
What is the purpose of Generative Models in unsupervised learning?
To model the underlying probability distribution of the data to generate new samples that look like the training data.
What is the role of the Generator in a GAN architecture?
To take random noise as input and transform it into data samples that look like real data.
How does the Discriminator contribute to the training process in a GAN?
It evaluates the authenticity of a given data sample and learns to distinguish between real and fake data.
What is the objective of the training process in a GAN?
To involve an adversarial game between the Generator and Discriminator to improve the generation of realistic data.
What is the significance of Convergence in a GAN?
It represents a point where the Generator produces high-quality data that is difficult for the Discriminator to differentiate.
What distinguishes a Deep Convolutional GAN (DCGAN) from other types of GANs?
DCGAN introduced convolutional layers to GANs, making them effective for image generation tasks.
What is Transductive Transfer Learning?
Transferring knowledge from a source task to a target task when the data distribution is the same but the objectives are different.
How does Unsupervised Transfer Learning work?
It transfers knowledge from a source domain with labeled data to a target domain with unlabeled data.
What is Feature Extraction in Transfer Learning?
It involves removing task-specific layers from a pre-trained model and adding new layers for the target task.
What is Fine-tuning in Transfer Learning?
Training the entire pre-trained model on the target task with a smaller learning rate.
How can Freeze and Unfreeze Layers be used in Transfer Learning?
Freezing certain layers preserves generic knowledge while unfreezing allows updates to those layers.
What are the Popular Pre-trained Models used for specific tasks?
ImageNet pre-trained models like VGG, ResNet, and GoogLeNet for visual recognition tasks like object detection.
How can Learning Rate Schedules benefit Fine-tuning?
Starting with a lower learning rate allows small adjustments before gradually increasing it.
What is the role of Dropout and Regularization in Transfer Learning?
Applying dropout and regularization techniques during fine-tuning to optimize performance.
How does Add New Task-Specific Layers contribute to Transfer Learning?
By adding new layers specific to the target task on top of pre-trained layers.
What are the key techniques in Fine-tuning for Transfer Learning?
Leveraging knowledge from pre-training, adapting it to the target task, and implementing learning rate schedules.
What is a filter in a convolutional neural network (CNN)?
A small matrix of weights that the network learns during training.
What is the purpose of a local receptive field in CNNs?
To break down complex visual information into smaller, manageable parts.
What is the feature map in a CNN?
Output matrix of the convolutional operation.
What is the main purpose of pooling layers in CNNs?
To perform spatial downsampling on the input feature maps.
What is the Flatten Layer used for in a neural network?
To convert the output of the previous layer into a one-dimensional array.
What is the purpose of fully connected layers in CNNs?
To connect the high-level features learned by convolutional and pooling layers to the final output layer.
What is the primary function of LeNet-5 architecture?
Handwritten digit recognition tasks.
Which CNN architecture achieved strong performance on image classification tasks and consisted of different configurations like VGG11, VGG13, VGG16, and VGG19?
VGG (Visual Geometry Group)
What is the purpose of YOLO (You Only Look Once) architecture?
Real-time object detection.
What are some common challenges in training CNNs?
Large datasets for effective training, computational intensity, sensitivity to biases, and vanishing/exploding gradients.
Learn about identifying interesting relationships in datasets and generating new data samples based on underlying probability distributions. Explore techniques like Apriori algorithm, GPT, and Generative Adversarial Networks (GANs).
Make Your Own Quizzes and Flashcards
Convert your notes into interactive study material.
Get started for free
