Machine learning:

Machine Learning is the study of algorithms that can "learn" or use past statistics and experiences to make prediction.

The definition of machine learning can go as follows:

"A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P if its performance at tasks in T, as measured by P, improves with experience E." -Tom M. Mitchell.

Machine Learning is very related to statistics, since it usually involves analyzing data to discover rules or patterns.

Common categories of machine learning include:

Supervised Learning: This is where exaples of imput/output pairs are given, and we want the machine to tell what the rule is that maps the inputs to these outputs.

Unsupervised Learning: Here only the inputs are given, the goal is to look for patterns.

Reinforcement learning: desired output is given. Environment is provided (without explicit description of the function, like in blackbox debugging). Input is requested.

There are many examples of categories of algorithms that "learn":

Artificial Neural Network (ANN): Complicated stuff

Deep Learning: Stuff about graph layers that I didn't get.

clustering (where you group object based on similarity in some aspect)

anomally detection

Applications of machine learning are very common nowadays. They include:

- recommendations made by search engines

- game playing

- machine perception

It can also be applied to:

- IoT

- wearable medical devices

IBM expects that in the next 5 years, machine learning will change the world completely:

- Classrooms will "learn" about the students throughout their education to ensure appropriate conditions and identify any problems.

- Local shops will "learn" about the local customers and be able to provide more customized products, beating online retailers once again.

- DNA analysis can be carried out much faster and will allow doctors to find the right treatment in hours or days instead of weeks and months.

- privacy will be protected using algorithms that learn about us and know whether we would be the ones who took such an action, instead of passwords that are prone to identity theft.

- cities will learn about their citozens and be able to provide better services specialized for them.

These predictions are very fascinating. I encourage everyone to watch the videos made by IBM under the name "5 in 5: Machine learning applications" which can be found here:



If all of these applications were made, what will we be spending most of our time doing?