In 2017, a learning computer is more than a reality. Machine Learning (ML) is a way of training an algorithm to learn without being explicitly programmed. It involves feeding huge amount of data into the algorithm and then allowing it to process that and learn. An example would be a translation algoritham that is fed a lot of translated material between sets of languages to help it improve.
Artificial Intelligence aims at creating machines that perform tasks that humans do. These could be general or narrow–meaning all human abilities or specific ones. AI can be achieved with or without machine learning.
These concepts have been around since 1940s and 1950s.The additional factor is the availability is huge amount of data. That makes a difference. Large organisations sit on data. The cloud is bringing computing power and ML is creating additional intelligence.
ML is helping oncologists through huge amount of cancer cases and suggest preferred treatment. Watson of IBM is an example. Entire cancer research can run into 50 million pages and 40,000 papers are added every year. It is impossible for a doctor to go through such huge data. ML is a great application to use in cancer treatment.The project has been implemented in Manipal Hospitals. It is an early adopter, possibly second or third hospital globally.
Arya.ai is working on creating an ML application for selling securities without letting the prices crash.
Algorthmic trading has been around for a while and brokers with proprietery trading arms use, it to gain a few seconds advantge. Research have focuses on whether an ML layer can be built on top of the algo. Can machines be allowed to alter the trading algorithm on their own? What will this mean to securities markets?
Google, Intel, Microsoft and Amazon have been developing off-the-shelf ML modules. The work is based on 1980s technologies of Jawa and Python. They have built their layers on top of that. Google has developed TensorFlow, Amazon AWS Machine Learning, Microsoft Azure Platform. In future, these modules will be available for non-experts too.
ML is useful in retail, transportation and financial services.
A radiologists job may become redundant in a few years. A radiologist interprets images. Millions of images (x-rays, CT scans, sonograms) can be fed and their interpretations too into an ML algorithm. A machine may give in due course a better interpretation than a human radiologist.
An ML application can sift through loan applications. There are bots to interface with the customers.
In India, there are hundreds of start ups using ML and AI technologies.
Deep Learning is a branch of machine learning that tries to mimic the structure of the human brain. Layers are created within the algorithm which pick specific parts of the subject to learn.