Unsupervised g
The method that the model explores hidden structure and images in unlabeled data.
Surveillance-free learning (unsupervised learning) is the paradig of a machine learning that the model works to explore any accurate response or label, only the hidden structure in the raw data, the image and relations itself. There is no "student" as in surveillance learning; the model creates its internal logic based on similarity and differences between data points. One of the most common unsupervised learning tasks is clustering (clustering): for example, K-means algorithm can separate customers to self-like groups looking at shopping behavior in a customer data set, without any predefined category.
Another important unsupervised learning task is to reduce size; techniques such as PCA (Basic Component Analysis) represent high-dimensional data at a lower-sized space by minimizing information loss, which facilitates visualization and accelerates subsequent modeling steps. Methods such as anomaly detection ( finding sapling data points from normal as in fraud detection) and autocoders (autoencoder). Un Surveillance learning is especially valuable when labeling cost is very high or practically impossible - for example, when it is asked to benefit from huge amount of unlabeled text or image data – especially if the pre-training phase of large language models is also an unsupervised (or self-supervised) learning process.
Another important unsupervised learning task is to reduce size; techniques such as PCA (Basic Component Analysis) represent high-dimensional data at a lower-sized space by minimizing information loss, which facilitates visualization and accelerates subsequent modeling steps. Methods such as anomaly detection ( finding sapling data points from normal as in fraud detection) and autocoders (autoencoder). Un Surveillance learning is especially valuable when labeling cost is very high or practically impossible - for example, when it is asked to benefit from huge amount of unlabeled text or image data – especially if the pre-training phase of large language models is also an unsupervised (or self-supervised) learning process.