Everything you need to know about Machine Learning –
Machine Learning is a branch of computer science that gives computer the ability to learn using statistical techniques given data without being explicitly programmed.
There are three types of Machine Learning:
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Supervised Learning:
This type of Machine learning needs input data that has labels. Let us take example of addition of two numbers. This type of learning is fast and accurate. First train the machine with data which is helpful in deriving the logic then test the logic using test data
Supervised learning are of two types:
- Classification
- Regression
Regression:
This type of technique is used for numerical data. For example, to find the price of house in specific house, regression is used.
Classification:
This type of technique is used for categorical data. This classifies the data into labels. For example, to detect whether a mail is spam or not, it is used.
UnSupervised Learning:
Unsupervised Learning takes data that do not have specific targets. There are two types of Unsupervised Learning. They are:
- Clustering
- Anomaly detection
Clustering:
This type of learning cluster items to different groups. For example, Classifying news paper articles into various groups.
Anomaly Detection:
This type of learning is used for detecting items in given data to detect items that does not fit to any category.
Reinforcement Learning:
This type of learning develops an algorithm that learns to react to environment. Mostly this type of learning is not used by the Data Analysts.
History of Machine Learning:
- In 1952, Arthur Samuel wrote the first ML algorithm to improve the game of the computer by learning winning moves from the people.
- In 1957, Rosenblatt designed the first neural network to simulate human brain thought process.
- In 1967, nearest neighbour algorithm is designed to discover the shortest path needed for travelers to cover a set of cities.
- In 1979, Students of Stanford University designed an algorithm to navigate obstacles in a room
- In 1997, IBM develops an algorithm to beat the world chess champion
- In 2006, ML is also used to see and identify objects
- In 2010, Microsoft developed an machine learning algorithm to detect human features
Recent Updates:
- In 2011, IBM develops an algorithm using ML to win Jeopardy competition
- In 2012, Google developed an algorithm to recognize cats on Youtube
- In 2014, Facebook develops an algorithm using Machine Learning for face recognition
- Lip Reading is another achievement in recent times
- In 2015, Google released Tensorflow framework using ML
- In 2016, AlphaGo developed an algorithm using ML to beat world chess champion
- In 2017, Google developed a Sonnet framework for ML
- In 2016, Facebook developed PyTorch framework based on Python for ML
Future Updates Machine learning:
- In future, ML involves Quantum Computing for faster processing.
- Also, Unsupervised Algorithm will be improved to provide better results.
- Using IoT with Machine Learning leads to generation of Edge Analytics that supports collaborative Learning.
- Also, the recommendations accuracy will be improved
- Some interesting things like Emotion detection, Voice recognition, Vision and facial recognition will be improved.
Connection with Artificial Intelligence & Deep Learning
Artificial Intelligence is a Computer science topic that has many subfields like Deep Learning, Computer Vision and ML etc. Artificial Intelligence is nothing but systems that are capable of getting solution to problems on their own.
Machine Learning means it ‘s not only provide solution to problem on their own, also learns and uses past knowledge to solve new problems. Deep Learning is a subset of ML that teaches computer to do things that are done naturally by human. Some the applications that are based on deep learning are driverless (Self-driving) cars, making cars to recognize a stop sign.
Software / Technologies:
- Apache Singa is an library that is used mainly for image recognition and Natural language processing.
- Shogun a software is written in C++ can be used on any operating system through various languages. It is used for implementing regression, classification algorithms.
- Apache Mahout is an open source library for implementing applications. Mahout can be used to implement filtering and classification algorithms.
- Apache Spark is an another ML library by Apache used for statistics, regression and classification. It is easy to use and scalable.
- Tensor Flow is Machine Learning library developed by Google used for applications that involve large number of numerical computations.
- Oryx is an library for Machine Learning for real time large scale machine learning applications. It is used to implement classification, filtering, regression and clustering algorithms.
- Accord.NET is a framework for Machine Learning used for scientific computing, pattern recognition, linear algebra, statistical analysis etc.
- Amazon Machine Learning is ML library developed by Amazon that has lot of visualization tools to create high end intelligent machine learning models.
- PredictionIO is an framework developed by Apache for Machine Learning which is mainly used for prediction algorithms.
- Eclipse Deep Learning is Machine Learning especially for Java which is used for developing deep learning models.
Free Resources:
- Microsoft Virtual Academy has some resources on using ML on Azure cloud.
- IBM provides better resources and tutorials to learn ML and Artificial Intelligence. The link to IBM resources is developer.ibm.com
- Lot of free courses on Machine Learning can be found here on Udemy.
Now, Please share this article on WhatsApp and Facebook, I’m sure they’ll love it!
And, please use comment section and let us know if you need any help or feel free to ask your questions.
