Machine Learning to HAR

Artificial Intelligence (AI) is the industry of this decade.

Machine learning (ML) is derived from Artificial Intellingence (AI).

Learning of machine is complex phenomenon division of computer science. At an abstract learning could be divided into supervised, semi-supervised and un-supervised learning.

Learning basically consitutes two stages i.e., training and validation.

For each division of learning there are different algorithms proposed over the years.

Normally, ML problems are either considered as

• Classification,
• Regression,
• clustering,
• Association rule mining,
• Structured output and
• Ranking.

I am currently working on a clasification problem, which belongs to supervised learning class of ML.

To solve classification problems, ML provides a great set of techniques/approaches. ML techniques could be further divided into old/traditional/classical ML techs and modern/recent deep-learing-based techs.

Fundamental difference between the old and modern ML techs is the automatic feature learning.

A big part of solving ML problems is in dealing with the dataset. Data could be in any format e.g. image, text, video, sound, sensor data etc. Broadly, data could be structured, semi-structured or unstructured.