Author: Anusree KJ
This blog helps you understand some basics about Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL). If you are completely new to the AI world, read this blog to understand some basics of AI, ML, and DL. If you are an expert in AI, skip this blog. Or if you want to read some interesting analysis in Football, read the blog on KNN Algorithm in Football Analysis.
In today's world, technology is growing very fast, with people developing new technologies day by day. One of the most used technologies of computer science in today’s life is Artificial Intelligence (AI). The main idea of AI is to make intelligent machines. Hence we can say AI is a study to make machines think like humans.
AI can be seen all around us now. For example, AI is mostly seen in a variety of subfields, ranging from general to specific, such as self-driving cars, playing chess, proving theorems, playing music, painting, etc.
Hence, AI is known as one of the fascinating & universal fields of Computer science and has a great scope in the future. AI makes a machine work just like how a human works.
Artificial Intelligence (AI)
Artificial Intelligence (AI) can make lots of difference to human intelligence and their life. AI helps machines behave more like humans in:
Talking
Thinking
Learning
Planning
Understanding
AI is also known as machine/Computer intelligence and is the result of the human need to perform complex calculations, and is achieved by high-speed computers. AI brings intelligence to machines.
The purpose of making machines intelligent is to make a person's life better, improve productivity, relieve people from complex tasks, and reach remote and dangerous places while keeping humans safe.
Machine Learning (ML)
Machine learning (ML) is a type of AI that allows software applications to predict outcomes accurately without the necessity of complex programming. Most ML algorithms use historical data as input to predict new output values.
Hence, we can say ML is a subset of AI, i.e., all ML algorithms are classified as being part of AI.
ML enables the machine to automatically learn from data, improve performance from past experiences, and make predictions. ML contains a set of algorithms that work on a huge amount of data. ML algorithms use these data to train, build the model and generate new output predictions from the built model.
Deep Learning (DL)
Deep Learning (DL) is a subset of ML. It was inspired by the working of human brains. The term Deep Learning usually refers to Deep Artificial Neural Networks.
Artificial Neural Networks (ANN) is a depiction of natural neurons in the brain. ANNs are mainly involved with deep learning algorithms and mimic the functionality of the human brain. The ANN can be used on all types of ML algorithms based on its functionality.
Hence, DL effectively teaches computers to do what humans naturally do, say learning by example. It uses Neural Networks (NNs) to extract higher-level data. NNs consist of 3 layers, say the input layer, the hidden layer, and the output layer. Each successive layer uses the output of the preceding layer as its input. For example, optical reading uses lower layers to identify edges and higher layers to identify letters.
ML and its types
There are many types of ML that can be used in different applications based on the type of datasets used. ML type is determined by different criteria like the type and amount of data, ML model requirement, how to train the ML model, etc.
Based on this, ML is divided into four types:
Supervised Learning
Unsupervised Learning
Semi-Supervised Learning
Reinforcement Learning
Supervised Machine Learning
Supervised machine learning is based on supervision. In this type of learning, we train the machines using the "labeled" dataset. Labeled data contains both input and output and is used to train the machines to learn from the data. Based on the trained model, the machine will predict output for new data.
More precisely, we can say; first, we train the machine with the input and corresponding output. Then we ask the machine to predict the output for new data using the trained model.
Unsupervised Machine Learning
Unsupervised learning is just the opposite of the Supervised learning technique. It does not require any supervision, i.e., the machine is trained using an unlabeled dataset, and the machine predicts the output without any supervision.
Unlabeled data does not have the output. It is a mixture of raw data, and the machine has to classify it based on some patterns.
For example, the unsupervised learning algorithm is given an input dataset containing images of different types of cats and dogs. The algorithm is not trained using the given dataset as it does have the output for corresponding input data. Hence it does not have any idea about the features of the dataset. The task of the unsupervised learning algorithm is to identify the images in the datasets on their own.
Here, the model performs some categorization by itself using some patterns identified from the input dataset. It will interpret raw data from hidden patterns, and then it will classify the data based on similarity using some suitable algorithms like k-means clustering, decision tree, etc.
The final output of the trained model will be a classified group of dogs and cats. This trained model can be later used to identify the new input images.
Semi-Supervised Machine Learning
Semi-Supervised learning is a mixture of both Supervised and Unsupervised machine learning. It represents the intermediate ground between Supervised and Unsupervised learning algorithms. It uses a combination of labeled and unlabeled datasets during the training period.
Google photo is an example of Semi-Supervised Learning. Google photos can show similar pictures based on the input name.
In the Semi-Supervised learning algorithm, training data is a combination of both labeled and unlabeled data. It contains only a small amount of labeled data and a large amount of unlabeled data. Initially, similar data is clustered using any unsupervised learning algorithm, and then it helps to label the unlabeled data into labeled data.
Below are some of the popular semi-supervised algorithms:
Deep Belled Networks
Restricted Boltzmann Machines
Reinforcement Learning
Reinforcement is another type of machine learning in which the system will learn by itself from the outcome of any action. Here, the system is exposed to a situation where it will take some action. Based on the action taken, it will be rewarded or penalized. The system then updates its policy with the actions it should and shouldn’t take. The task of the system is to get rewarded and not get penalized.
Reinforcement learning works on a feedback-based process in which an AI agent (A software component) automatically explores its surroundings by hitting & a trail, taking action, learning from experiences, and improving its performance. Agents get rewarded for each good action and get punished for each bad action.
In reinforcement learning, there is no labeled data like supervised learning, and agents learn from their experiences only. This type of learning process is similar to a human being; for example, a child learns various things through experiences in his day-to-day life.
We don't have to do any programming for this type of learning as it learns from its own experience.
Algorithms for different types of Datasets
Supervised Learning and its Types
1. Classification
The Classification algorithm is a Supervised Learning technique that is used to solve classification problems in which the output variable is categorical, such as "Yes" or No, Male or Female, etc. These algorithms predict the categories present in any dataset. Spam Detection, Email filtering, etc., are some examples of classification problems.
Following are some popular Classification algorithms that come under supervised learning:
KNN
Random Forest
Decision Trees
Logistic Regression
Support vector Machines
2. Regression
Regression is a technique used to solve regression problems. That is, it is used for investigating a linear relationship between independent variables or features and a dependent variable or outcome.
Regression is predictive modeling in machine learning, in which an algorithm is used to predict continuous outcomes. They are used to predict continuous output variables, such as market trends, weather predictions, etc.
Below are some of the Regression algorithms under supervised learning:
Linear Regression
Regression Trees
Non-Linear Regression
Bayesian Linear Regression
Polynomial Regression
Unsupervised Learning and its Types
1. Clustering
Clustering is a method of grouping objects into clusters such that objects with more similarities remain in one group.
2. Association
An association is used for finding the relationships between variables in a large database. Association rule makes marketing strategy more effective and easy to operate.
For example, people who buy bread also tend to purchase Butter/Jam.
Below are some of the unsupervised learning algorithms:
K-means clustering
KNN (k-nearest neighbors)
Hierarchical clustering
Anomaly detection
Neural Networks
Principal Component Analysis
Independent Component Analysis
Apriori algorithm
Singular value decomposition
Conclusion
Artificial Intelligence (AI)), also known as machine/Computer intelligence, is the result of the human need to perform complex calculations and is achieved by high-speed computers. AI brings intelligence to machines.
Machine Learning (ML) is a subset of AI that allows software applications to predict outcomes accurately without the necessity of complex programming.
Deep Learning (DL) is a subset of ML. It effectively teaches computers to do what humans naturally do. DL mimics the functionality of the human brain. It uses neural networks to predict the output.
Based on the type of data and tasks, Machine learning is classified into four types:
Supervised learning
Unsupervised learning
Semi-supervised learning
Reinforcement Learning
Supervised learning only uses labeled data for training the model. Labeled data will have both the input and output and is fed to the model for training.
Unsupervised learning, on the other hand, is used for training unlabeled or raw data without output. Output is not known before training, and therefore the model has to cluster the data or group it based on the similarity of selected parameters.
Semi-Supervised learning is a mixture of supervised learning and unsupervised learning, as it uses both labeled and unlabeled data. It uses a few labeled and more unlabeled data.
Reinforcement Learning is much different. Here the system is exposed to a situation where it can take action. According to the action taken, the system will be rewarded or penalized. Tasks of the system will be to avoid getting penalized.
Continue reading about the KNN algorithm and how it is suitable to find the 10 best player recommendations for a replacement of any player in Football in KNN Algorithm in Football Analysis.
If you are new to Machine learning, then go through A Basics to KNN algorithm before moving on to KNN Algorithm in Football Analysis.