Unsupervised learning means to "explain" complex or unknown data in a way that is objective, independent yet relevant.
For example, shop owners try always to learn new ways of increasing the profit by watching the passers-by in search for new ideas – an idea such as street attraction during vacation times, a need for a quick service in rush hours, offering help to window shoppers, etc. A fresh and objective look on unfiltered data is the key for useful indications that can be acted upon.
Many practical ideas can be found methodologically by analyzing available unsupervised data.
Unlike supervised-learning that pre-assumes much of the information about variables, in unsupervised learning one acquires new understanding regarding variables as well as cluster formation, alerts and farther insights.
Unsupervised learning is a type of machine learning where an algorithm learns patterns and relationships in data without being explicitly trained or labeled by a human. In unsupervised learning, the algorithm is given a set of input data and tasked with finding meaningful structures or patterns in the data on its own.
The goal of unsupervised learning is to uncover hidden patterns and structures in the data, which can be useful for tasks such as data clustering, anomaly detection, and dimensionality reduction. Unlike supervised learning, where the algorithm is provided with labeled training data, unsupervised learning is more exploratory in nature and is used when the desired output is not known in advance.
Unsupervised learning algorithms can be broadly categorized into two types: clustering and association. Clustering algorithms group similar data points together based on their similarity, while association algorithms identify patterns of co-occurrence or association between different features in the data.
Examples of unsupervised learning algorithms include k-means clustering, principal component analysis (PCA), and autoencoders. These algorithms are widely used in fields such as computer vision, natural language processing, and recommendation systems.
Dear Edith Ohri ,
What is unsupervised learning? Unsupervised learning, also known as unsupervised machine learning, uses machine learning algorithms to analyze and cluster unlabeled datasets. These algorithms discover hidden patterns or data groupings without the need for human intervention.
https://www.ibm.com/topics/unsupervised-learning#:~:text=the%20next%20step-,What%20is%20unsupervised%20learning%3F,the%20need%20for%20human%20intervention.
Regards,
Shafagat
Unsupervised learning is a type of machine learning that looks for previously undetected patterns in a data set with no pre-existing labels and with a minimum of human supervision.
Unsupervised learning is when you have a data set and apply an algorithm to find latent patterns in data. This algorithm contains a mathematical/statistical formula. In other words, apply this formula to some data, check the results, and see if you can make sens of the findings.
Dear Shafagat Mahmudova, thanks for your answer. It is indeed the widely accepted definition. The idea behind the question was to go one more step and see what makes the difference between supervised and unsupervised learning. It seems to me that the core issue here has to do with the nature of data that is now large, fast, and multi-sourced. Data has become unsupervised to a degree that analysts find it hard even to formulate a good goal function...
Dear Arnab D.
Appreciate your thorough answers and explaining the mechanism of unsupervised learning.
Let me just comment that learning without supervision, labels and specific target variable is more an ideal than a real working model. It can be proven for instance, that clusters are a function of the target. Not having a specific target should than produce an empty set of clusters. Also, in order to have "meaningful results" one need to define some kind of a target. I think that this conundrum has a solution, but not the one that is in the books :)
Muhamed Hassen Seid. Your point is important. Much agree that a learning model should detect new patterns. Discovery is the whole story...
Daniel Dan, generally it's tight that.unsupervised learning uses algorithms to find hidden pattern.
Let me use this conversation to clear a shaky practice that is expressed in the answers - the reliance on algorithms as if they can magically produce good results. In fact mathematics and statistics are just theoretical tools that have nothing to say about the real world. The actual algorithm (that may apply them) is up to us the analysts, and we are judged at the end of the day by the ability to add new knowledge.
Edith Ohri, mathematics and the practical applications in statistics offer models that need interpretation after their application. With their help, we are trying to get latent dimensions in data. What I said is correct, see it as a crasis in linguistics, just to make a comparative procedure.
Daniel Dan True yet all models have conditions, which in the case of unsupervised data are hard to satisfy.
What is the purpose of an analysis model? In my view, it is to offer a popular solution with a small scale option for individual end users. In contrary,If it needs "talent", long & expensive training or high qualification analysts, it misses the point. My solution GT data mining found a simpler path... Sorry that the project "new data science" is no longer available at ResearchGate.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs