4.1 Artificial Intelligent Algorithms
Artificial intelligence algorithms are generally grouped into three categories. These are Supervised Learning, Unsupervised Learning, and Reinforcement Learning.
Unsupervised Learning case, target output is not given, and the model is expected to form a template from the given inputs.
Reinforcement Learning is a special case of supervised learning. This learning case, you are give a rating that specifies how accurate the output of the model is.
In the supervised learning method, the model is fed with different data and the results of the model are compared with the actual target outputs. The errors are tried to be minimized by the model.
Supervised learning is generally examined in two heading which are Classification and Regression.
Classification case example can be sentiment analysis. For example, "this hotel room can be considered the worst room to stay" → negative. "Good service, good quality service, beautiful landscape" → is regarded as positive. The results have already appeared negative or positive and it is observed which output will be taken.
In regression, output is the preferred method in cases where output is generated from continuous numbers rather than negative, positive definitions. Regression can be a field of use where professions have a score s based on the results inputs.
Unsupervised learning is often used in finding patterns in inputs. The clustering algorithms are under the heading of unsupervised learning.
Clustering algorithms can be used under both supervised and unsupervised methods in situations such as anomaly detection (awareness of abnormal situations). Likewise , artificial neural networks are also under the heading of regression and clustering.
Selection of the algorithms for the model is the most important question. To provide a general understanding of this situation, the map prepared by scikit-learn.org gives a general idea of classification, clustering, regression and dimension reduction.
While many different algorithms continue to be developed, generally accepted ones can be grouped under the following headings.
Simple Regression
Decision Tree
SVM (Support Vector Machine)
Bayesian Classifiers
Clustering
PCA (Principle Component Analysis)
Deep Learning
Besides the above algorithms, many algorithms are used in the artificial intelligence. We gave some of the generally accepted titles in this post. We will use algorithms one by one in our following writing.
With our deduction method, you can look again at the link below to remember our general goal.
