The process of adding a penalty to the loss function to discourage complex models is called ________.
- Normalization
- Optimization
- Parameterization
- Regularization
Regularization is a technique used in machine learning to add a penalty to the loss function, discouraging overly complex models and preventing overfitting. It helps improve a model's generalization to new data.
What is the central idea behind using autoencoders for anomaly detection in data?
- Autoencoders learn a compressed data representation
- Autoencoders are trained on anomalies
- Autoencoders are rule-based
- Autoencoders use labeled data
Autoencoders for anomaly detection learn a compressed representation of normal data, and anomalies can be detected when the reconstruction error is high.
In convolutional neural networks, using weights from models trained on large datasets like ImageNet as a starting point for training on a new task is an application of ________.
- Transfer Learning
- Regularization
- Batch Normalization
- Data Augmentation
This application of transfer learning involves using pre-trained CNN models, like those on ImageNet, to initialize weights in a new model for a different task. It accelerates training and leverages existing knowledge.
While LSTMs have three gates, the GRU simplifies the model by using only ________ gates.
- 1
- 2
- 3
- 4
Gated Recurrent Units (GRUs) simplify the model by using only two gates: an update gate and a reset gate, as opposed to the three gates in LSTMs.
In a situation where you have both numerical and categorical data, which clustering method might pose challenges, and why?
- Agglomerative Clustering
- DBSCAN Clustering
- Hierarchical Clustering
- K-Means Clustering
K-Means may pose challenges in such a situation because it calculates centroids using the mean, which isn't well-defined for categorical data. Other methods like hierarchical or DBSCAN may be more suitable.
An online retailer wants to create a hierarchical structure of product categories based on product descriptions and features. They want this hierarchy to be easily interpretable and visual. Which clustering approach would be most suitable?
- Hierarchical Clustering
- DBSCAN
- Gaussian Mixture Model (GMM)
- Affinity Propagation
For creating a hierarchical structure, Hierarchical Clustering is the most suitable approach. It builds a tree-like structure that is interpretable and can be easily visualized. This makes it ideal for organizing product categories based on descriptions and features.
Experience replay, often used in DQNs, helps in stabilizing the learning by doing what?
- Reducing Correlation between Data
- Speeding up convergence
- Improving Exploration
- Saving Memory Space
Experience replay in DQNs reduces the correlation between consecutive data samples, which stabilizes learning by providing uncorrelated transitions for training.
Time series forecasting is crucial in fields like finance and meteorology because it helps in predicting stock prices and ________ respectively.
- Temperature
- Rainfall
- Crop yields
- Wind speed
Time series forecasting in meteorology is important for predicting variables like rainfall, not stock prices.
In the context of Q-learning, what does the 'Q' stand for?
- Quality
- Quantity
- Question
- Quotient
In Q-learning, the 'Q' stands for Quality, representing the quality or expected return of taking a specific action in a given state.
Which regression technique uses the logistic function (or sigmoid function) to squeeze the output between 0 and 1?
- Linear Regression
- Logistic Regression
- Poisson Regression
- Ridge Regression
Logistic Regression uses the logistic function (sigmoid function) to model the probability of a binary outcome. This function ensures that the output is constrained between 0 and 1, making it suitable for classification tasks.