Classification of Machine Learning

According to the type of learning "signal" or "response" that a learning system is able to receive, machine learning implementations can be divided into four broad groups, which are as follows:

1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement learning
4. Semi-supervised learning

When one considers the desired output of a machine-learned system, another classification of machine learning tasks emerges:

1. Classification
2. Regression
3. Clustering

Supervised Learning

As the name suggests, supervised learning involves a supervisor serving as an instructor. In its simplest form, supervised learning refers to the process of teaching or training a computer system using labelled data. Which indicates that the right answer has already been assigned to certain data. In order for the supervised learning algorithm to analyze the training data (set of training examples) and provide an accurate result from labelled data, the machine is then given a fresh set of examples (data).

Consider the scenario when you receive a basket full of various fruits. The first step now is to teach the machine each variety fruit individually. The label "Apple" will be present if the item is rounded, has a depression at the top, and is red in color. If an object has a long, curved cylinder shape and the hue green-yellow, it will be given the name "-Banana." Now imagine that you have provided a new, distinct fruit—say let's a banana from the basket—and asked someone to identify it after training the data.

As a result of earlier data, the machine has already learnt certain things, therefore this time it must be used carefully. It will first categorise the fruit based on its shape and colour, confirming that it is a BANANA and placing it in the category of Bananas. As a result, the computer gains expertise from training data (a fruit basket) and applies it to test data (new fruit).

Two groups of algorithms are used in supervised learning:

Classification: When the output variable is a category, such as "Red" or "blue," "disease" or "no disease," a classification problem exists.
Regression: When the output variable has a real value, such "dollars" or "weight," a regression problem exists. With "labelled" data, supervised learning works with or learns. This suggests that some data has already been assigned the right response.
Types:

1. Regression
2. Logistic Regression
3. Classification
4. Naive Bayes Classifiers
5. K-NN (k nearest neighbors)
6. Decision Trees
7. Support Vector Machine

Unsupervised Learning

Unsupervised learning is the process of teaching a computer to use unlabeled, unclassified data and allowing the algorithm to act on the data without supervision. Without any prior data training, the machine's objective in this case is to categorise unsorted data according to similarities, patterns, and differences. In contrast to supervised learning, no teacher is present, hence the machine won't be trained. As a result, the machine is limited in its ability to discover the hidden structure in unlabeled data on its own.
Consider the scenario where it is shown a picture of dogs and cats that it has never seen before. Therefore, we cannot classify the machine as "dogs and cats" because it is unaware of the characteristics of dogs and cats. However, it can divide them into groups based on their patterns, similarities, and differences, thus we can simply divide the image above into two halves. The first section might have only images with dogs, and the second segment might have just images with cats. You have no prior knowledge of this, hence there are no examples or training data.
It enables the model to function independently and find previously unnoticed patterns and information. It focuses primarily on unlabeled data.

Unsupervised learning is divided into two groups by the algorithms they fall under:

Identifying the natural groupings in the data, such as classifying clients based on their purchasing patterns, is a clustering problem. When you wish to find rules that broadly characterise your data, such as "those who buy X also tend to buy Y," you have an association rule learning problem. Various Unsupervised Learning Methods are: Clustering

1. Exclusive (partitioning)
2. Agglomerative
3. Overlapping
4. Probabilistic
   

Clustering Types

1. Hierarchical clustering
2. K-means clustering
3. Principal Component Analysis
4. Singular Value Decomposition
5. Independent Component Analysis

Reinforcement Learning

Machine learning includes the discipline of reinforcement learning. It involves acting appropriately to maximize reward in a certain circumstance. It is used by a variety of programs and machines to determine the optimal course of action to pursue in a given circumstance. There is no correct answer in reinforcement learning, but the reinforcement agent selects what to do to complete the job. This is different from supervised learning, where the training data includes the solution key and the model is trained with that answer. It is obligated to gain knowledge from its experience in the absence of a training dataset.

1. Input: The input should be an initial state from which the model will start
2. Output: There are many possible outputs as there are a variety of solutions to a particular problem
3. Training: The training is based upon the input, The model will return a state and the user will decide to reward or punish the model based on its output. The model keeps continues to learn. The best solution is decided based on the maximum reward.
   

There are two distinct categories of reinforcement:

1. Positive: Reinforcement is defined as when an event, occurs due to a particular behavior, increases the strength and the frequency of the behavior. In other words, it influences behavior in a favorable way. The following benefits of reinforcement learning: Boosts Performance and Maintain Change for a Significant Period The results may be weakened by an excess of states brought on by excessive reinforcement.
2. Negative: Negative Reinforcement is defined as strengthening of behavior because a negative condition is stopped or avoided. Advantages of reinforcement learning: Increases Behavior Provide defiance to a minimum standard of performance and It Only provides enough to meet up the minimum behavior.

Semi-Supervised Learning

When a training set with some (typically many) of the target outputs missing is provided, this is referred to as an incomplete training signal. An exception to this rule known as Transduction occurs when the whole set of issue cases is known at the time of learning but only a portion of the targets are present. Semi-supervised learning is a method of machine learning that, during training, blends a sizable amount of unlabeled data with a small amount of labelled data. Between supervised and unsupervised learning is semi-supervised learning.

Classification

The learner must create a model that categorises unseen inputs into one or more of the classes when inputs are divided into two or more groups (multi-label categorization). Usually, this is handled under supervision. When email (or other) messages are the inputs and the classifications are "spam" and "not spam," that is an example of categorization.

Regression

Which is also a supervised problem, A case when the outputs are continuous rather than discrete.

Clustering

When a set of inputs is to be divided into groups. Unlike in classification, the groups are not known beforehand, making this typically an unsupervised task.

In this blog, we just had an overview on the different classification of Machine Learning, and in the upcoming blogs we will cover Data and its processing, Data cleaning, How data is processed, etc. Thanks for reading, Have a Great Day !