Machine Learning
It is the ability to perform the tasks which comes naturally to human beings and capability to learn without being explicitly programmed. It is about studying from data, identifying patterns and to make decisions. The growth of artificial intelligence has been assigned to growing role of data, computational processing and affordable data storage with which it is possible to quickly and automatically produce models that can analyze complex data and deliver faster and more accurate results even on a very large scale.
It have applications in understanding transportation patterns and trends, retail services, oil and gas like analyzing the minerals, internet of things, financial services like preventing fraud and will also have application in the healthcare like wearable device.
Machine learning and data mining have some interrelation which is basically that data mining uses different methods to extract insights from it, it also involves data storage and manipulation which is a tool for transforming information into knowledge, with the explosion of data the need that the hidden pattern in it should be established emerged, so that future events can be predicted like detecting cancer, self-driven cars, creating new drugs. It uses information and answers to discover the rules behind a problem, it tries different rules and learns how well they are performing.
Types of Machine Learning
- Supervised – It is used in case of labelled data where desired output is known, for example temperatures for the different days and number of beach visitors on a particular day.
- Unsupervised – It is used in case of unlabelled data where desired output is not known.
- Semi Supervised – It is a combination of supervised and unsupervised.
- Reinforcement – It is based on trial and error where the correct model is rewarded. It is used in gaming, navigation, reinforcement is used for experiential studying in artificial intelligence where the correct moves are recorded, for example-Alpha Go an AI system developed by Deep mind a subsidiary of Google.
Deep Learning
It is a cherished goal of artificial intelligence which is based on more complex artificial neural networks called deep neural network. It is a combination of advanced computing and special types of neural network.
For example behavioural neural network used in NEON( Neo + Human) called virtual human developed by Starlabs of Samsung. These behavioural neural networks are performing 3Rs i.e Reality, Real time and Responsiveness.
It is about using brain re to learn from large information and will get desired push from the development of neuromorphic computing which is based on the complexity of the brain architecture.
Recently University of Manchester developed Spinnaker, a neuromorphic supercomputer that is Spiking Neural Networks Architecture it is a replication of the spikes of neurons. This computer has used the research from the human brain project of European Union involving neuroscientist and computer scientist. The ultimate goal is to develop the critical thinking ability in Deep learning where the performance keeps on improving with the increase in the data quantity. It will be used in edge computing which will be used in smartphones, car, robots to give the desired push to artificial intelligence.
Conclusion
We can conclude that Artificial Intelligence has evolved into Machine and Deep Learning. Machine studting is the ability to perform task which humans can do naturally such as identifying patterns, decision making, etc. There are different types of Machine Learning such as Supervised, Unsupervised, Semi-supervised and Reinforcement and Deep Learning is the sub part of it which is more complex and is a combination of advanced computing and special types of neural network.
FAQs
What exactly machine learning?
It is a ability to perform the tasks and capability to learn without being explicitly programmed. Studying from data, identifying patterns and to make decisions.
What are the 3 types of machine learning?
- Supervised – It is used in case of labelled data.
- Unsupervised – It is used in case of unlabelled data.
- Reinforcement – It is based on trial and error and the correct model is rewarded.
What is machine learning with example?
Tasks are performed by the machine which is generally done my human beings example navigation, gaming, calculating temperatures for different days.
