What is the module for detecting outliers in Torch?

In Torch, outlier detection is typically achieved through statistical methods or machine learning-based models. Here are some commonly used outlier detection modules.

1. One-Class SVM is an unsupervised learning algorithm used for detecting anomalies in data by trying to find a boundary that separates the majority of data points (normal points) to identify the outliers.
2. Isolation Forest is an anomaly detection algorithm based on decision trees. It constructs a group of isolated decision trees by randomly selecting features and partitioning data, effectively identifying outlier points.
3. Autoencoders are a type of neural network model that can be utilized for anomaly detection. Training an autoencoder to reconstruct input data often results in higher reconstruction errors for anomalies, making them detectable.
4. Statistical methods, such as Z-Score, IQR (Interquartile Range), and Grubbs Test, based on statistical principles are used to detect outliers by calculating and comparing statistical properties of the data.

In Torch, you can implement these outlier detection methods using related libraries or modules, such as using torch, scikit-learn, or other third-party libraries to implement the required outlier detection algorithm.