what is machine learning, what are the types of machine learning, supervised and unsupervised learning
What is Machine Learning?
Machine Learning is a type of artificial intelligence that focuses on the development
of computer programs that use computer systems that can learn and adapt
without being explicit instructions, by using algorithms and complex models
to analyze and draw a conclusion from patterns in data. It is a subset of
Artificial Intelligence.
Types of Machine learning
Machine learning can be categorized as:
1. Supervised Learning
Supervised machine learning is where involves variables that mapping function from
an input to an output based on a series of pairs.
For example, if we have a data set of two variables one being symptoms
which is the input, and the other being the problems as output. And
implement a supervised learning model to predict the problem of a person
based on their symptoms. Further, with supervised learning, there are two
subcategories one is regression and the other is classification.
1.1 Regression Model
In the regression model, we find a target value based on independent
predictors. That means you can use this to find the relationship between a
dependent variable and an independent variable. In regression models, there
is a continuous output.
Use cases:
·
Risk Assessment
·
Score Prediction
·
Weather Forecasting
·
Population Growth Prediction
·
Market Forecasting
Some of the common types of regression models include
Linear regression
is simply finding a line that fits the data its extensions include multiple linear regression that is finding a plane of best fit and polynomial
regression that is finding a curve for best fit.
The Decision tree
looks something like this where each square above is called a node and the
more nodes you have the more accurate your decision tree is.
Random forest
these are assembled machine learning techniques used for a variety of tasks
that builds off over decision trees and involve creating multiple decision
trees using original data and randomly selecting sub-samples of variables.
Then the model predictions the output of the decision tree.
A Neural network
is quite popular and is a multi-layered model inspired by human minds like
the neurons in our brain the circle represents a node the blue circle
represents an input layer the black circle represents a hidden layer and the
green circle represents the output layer each node in the hidden layer
represents a function that input goes through ultimately leading to the
output in the green circles.
1.2 Classification
It is a process of categorizing a given data or observations into several
classes, which can be structured/ unstructured data. It attempts to predict
the class of given input data points.
Use cases:
·
Fraud Detection
·
Email Spam Detection
·
Image Classification
·
Diagnostics
·
Identify Fraud Detection
In classification, there is a discrete output and some of the
classification models include:
Logistic regression
is like linear regression but is used to model the probability of a
finite number of a binary outcome.
Support vector machine is a supervised classification technique that carries a goal to find
a hyper lane in n-dimensional space that can distinctly classify the data
points.
The Naive Bayes
machine learning model is a probabilistic classifier used for a variety of classification tasks
and works on the Bayes theorem.
Decision trees A decision tree algorithm is a type of supervised learning. It is used for classification and prediction. It shows the sequence and the structure of a decision problem.
2. Unsupervised Learning
Unlike supervised machine learning, unsupervised learning is used to draw a
conclusion and find patterns from input data sets without references to the
labeled outcome.
Two main methods used in supervised learning include:
Clustering
Clustering involves the grouping of data points. It is frequently used for
customer segmentation, fraud detection, and document classification. Common
clustering techniques include k-means clustering, hierarchical clustering
means shape clustering, and density-based clustering. While each technique
has different methods in finding clusters they all aim to achieve the same
thing.
Use cases:
·
Targeted Marketing
·
Customer Segmentation
·
Recommender Systems
Dimensionality reduction
It is a process of reducing dimensions of your feature set by reducing the number of features most dimensionality reduction techniques can
be categorized as either feature elimination or feature extraction a popular
method of dimensionality reduction is called principal component analysis or
PCA.
Use cases:
·
Face Recognition
·
Image Recognition
·
Text Mining
·
Big Data Visualization
·
Meaningful Compression
3. Semi-supervised Learning
Semi-supervised machine learning uses a combination of supervised being fully
labeled and unsupervised learning being unlabeled data.
For example, accessing a large unlabeled data and training a model,
manually labeling all the states is not practical. Manually label some
part of this large data set and use that part to train our model. The
remaining data set we have to train a model and more robust our model will
be. One can do is implement a technique in the category of semi-supervised
learning called pseudo labeling.
This is how pseudo labeling works, labeled some part of our data set to
use this label data as the training set for the model. Then train our model
with any other labeled data. Then regular training process to get our model.
The unsupervised learning piece comes into play. After training our model in
the labeled part of the data set, predict the remaining unlabeled data
using this model. Label every piece of unlabeled data that is predicted with
the individual outputs.
Train our model on the full data set which is now comprised of both the
data labeled along with the pseudo labeled data. Through the use of pseudo
labeling to train on a larger data set we're also able to train on data that
otherwise potentially taken many hours of human labor to label the data.
The act of generating all the labels itself is not workable. Through
this process, one can make use of both supervised learnings with the labeled
data and unsupervised learning with the unlabeled data which together give
us the practice of semi-supervised.
4. Reinforcement learning
Reinforcement machine learning falls under artificial intelligence it involves
topics like goal setting, planning, and perception. It can even form a
bridge between AI and the engineering disciplines. A Self-learning system
relies on reinforcement learning. Several companies start to public
reinforcement learning in the recommendation and personalized systems.
Use cases:
·
Finance Sector
·
Inventory Management
·
Game AI
·
Real-time decisions
·
Skill Acquisition
·
Robot Navigation
COMMENTS