Deep Learning Technical Q&A

Which deep learning framework provides tools for reinforcement learning tasks in games?
Answer: PyTorch.

What is "policy gradient" in the context of reinforcement learning?
Answer: A technique for training agents to play games using neural networks.

Which reinforcement learning algorithm involves updating the Q-values of state-action pairs based on temporal difference errors?
Answer: Deep Q-Network (DQN).

What is a neural network in the context of machine learning?
Answer: A computational model inspired by the human brain's structure.

What is the basic building block of a neural network?
Answer: Neuron (or Node).

What is the primary advantage of using deep neural networks (DNNs)?
Answer: DNNs can automatically learn hierarchical features.

What is "transfer learning" in the context of machine learning?
Answer: The process of reusing pre-trained models on a new but related task.

What is the main advantage of transfer learning?
Answer: It reduces the need for labeled data in every new task.

What is "fine-tuning" in transfer learning?
Answer: A process for training a model on a new task while adjusting some of the pre-trained layers.

Which neural network architecture is commonly used for transfer learning?
Answer: Convolutional Neural Networks (CNNs).

What is "AutoML"?
Answer: The use of automated techniques to perform various tasks in the machine learning pipeline.

What is the goal of AutoML?
Answer: To automate the entire machine learning process, from data preparation to model deployment.

Which deep learning architecture is commonly used for image classification tasks?
Answer: Convolutional Neural Networks (CNNs).

What is the main advantage of using CNNs for image classification?
Answer: CNNs can automatically learn hierarchical features from images.

Which deep learning framework provides pre-trained models and tools for image classification tasks?
Answer: TensorFlow.

What is "data augmentation" in the context of image classification?
Answer: The process of generating new data samples from existing data using various transformations.

What is the primary goal of object detection in deep learning?
Answer: Detecting and localizing objects within an image.

Which deep learning architecture is commonly used for object detection tasks?
Answer: Convolutional Neural Networks (CNNs).

What is the "anchor box" concept in object detection?
Answer: A predefined box used for localizing objects of different scales and aspect ratios.

Which deep learning architecture is commonly used for semantic segmentation tasks?
Answer: Convolutional Neural Networks (CNNs).

Which neural network architecture is designed for sequential data and natural language processing tasks?
Answer: Recurrent Neural Network (RNN).

What is the primary advantage of using Long Short-Term Memory (LSTM) networks over traditional Recurrent Neural Networks (RNNs)?
Answer: LSTMs can capture long-range dependencies in sequences.