Artificial Intelligence

Artificial Intelligence (AI) is one of the emerging technologies growing exponentially and Forrester Research predicted a greater than 300% increase in investment in artificial intelligence in 2017 compared with 2016. IDC estimated that the AI market will grow from $8 billion in 2016 to more than $47 billion in 2020. This article is written to understand AI especially for beginners and for developers who are inclined to develop software AI machines. Topics covered are as below:

  1. What is Artificial Intelligence?
  2. Evolution of Artificial Intelligence
  3. Examples of AI implementations
  4. Different areas of Artificial Intelligence
  5. how deep learning works
  6. Steps to transform IT operations with Artificial intelligence
  7. List of few Artificial Intelligence software
  8. AI technologies


What is Artificial Intelligence?

Humans are considered most intelligent creatures and can learn, think and make intelligent decisions. Artificial Intelligence (AI) is a branch of Computer science and goal is to create computer systems and machines that can Learn and act like humans. However, as humans, we have a limitation to analyse complicated multi-dimensional problems and problems involving huge data. With artificial intelligence we can programme a machine to mimic human behaviour, to learn, to predict, to take decisions and to solve problems involving huge data with multi-dimensional parameters.

Evolution of Artificial Intelligence

AI has been there since 1950’s and the concept was coined in 1950’s at a Dartmouth conference and in initial projects in 1960’s were McCulloch & Pitts Boolean model of brain, Turing Test by Alan Turing. In 1960’s Space Odyssey film features the sentient of deadly computer. In 1980’s Neural Networks and machine learning became popular. The concept was visualized in the movie “Terminator”. Using the AI technology in 1990’s, IBM deep blue computer beats world chess champion Gary Kasparov. So we have had AI theoretical models for a long time but we see it practically advances exponentially during recent few years and this has become possible due to the increase in large data and high computing power.

Examples of AI implementations include

  • IBM Watson wins Jeopardy
  • Google Brain computer clusters and trains itself
  • Siri – acts as a personal assistant using voice processing
  • Facebook – automatically tags the people using image processing
  • Amazon- recommends products to purchase using machine learning
  • Cortana is a virtual assistant created by Microsoft for Windows 10
  • Alexa is an intelligent personal assistant developed by Amazon
  • Cyc is the world’s longest-lived artificial intelligence project, attempting to assemble a comprehensive ontology and knowledge base that spans the basic concepts and “rules of thumb” about how the world works
  • AlphaGo is a computer program that plays the board game Go
  • DeepDream is a computer vision program which uses a convolutional neural network to find and enhance patterns in images thus creating a dream-like hallucinogenic appearance in the deliberately over-processed images.

Different areas of Artificial Intelligence

There are different categories of artificial intelligence as shown in below figure. Machine learning is a way to train the machine and it learns from data provided and rules coded in the program. Deep learning is a subset of machine learning and it has algorithms built on a model similar to the way human brain works, described in next section.

Figure 1- Classification of Artificial Intelligence fields

Image Processing – Machines can be trained to recognize objects

Computer Vision – Machines can see and process what they see

Robotics – Machines can learn the soundings and move intelligently

Speech Recognition – Machines can speak, Listen and communicate like humans. Speech recognition transforms human speech into format useful for computer applications. Especially used in interactive voice response systems and mobile applications.

Pattern recognition – Have the ability to form patterns and grouping

Natural Language Processing – Machines can read, write and translate text in any language. Primarily Producing text from computer data. It is used for text analytics to understand the sentence and intent through statistical and machine learning methods. Example Expert systems, Mindbreeze etc

Machine Learning – We enable computers to learn without actually writing a program unlike what is done for other applications. We have to train the machine on a large dataset to create a model which helps the machine to take decisions based on its learning. For example if we have to train a computer to recognize a pen from a pencil. We need to train the machine by providing a large dataset that contains various attributes of pens and pencils. The dataset will consist of type of nib or lead and other attributes that differentiate a pen from pencil. Using this dataset, the machine will create a model which can be used to differentiate a pen from pencil. So next time whenever we pass object (pen/pencil) attributed to the model, it will give the result – pen or pencil. Below is a depiction of traditional applications vs. machine learning.

Figure 2- Traditional Application vs. Artificial intelligence

Machine Learning is further classified under three categories

  • Supervised Learning – If we train a machine with data and algorithm that has the answer as well, example tag the faces of people with their names.
  • Unsupervised Learning – If we train the machine with large dataset and want the machine to derive a pattern and learn by itself.
  • Reinforcement Learning – If the machine learns through trial and error method

Deep Learning – A subset of machine learning consisting of artificial neural networks with multiple abstraction layers. Especially used in grouping, pattern recognition and classification applications. In deep learning we teach a computer to solve problems like humans. We have to show the machine how the problem was solved previously and it learns the steps like the way human brain learns.

Let’s see how deep learning exactly works like human brain.

Our brain is made up of zillions of neurons connected to form a neural network. These neurons communicate with each other using dendrites as in the below figure. The cell body processes the information and sends information the connecting neurons. Each time information is passed from one cell to the other, the information becomes clearer and the brain learns it and the last layer is output layer.

Similarly in deep learning, the model is like human brain with artificial neural network, with layers and each layer connected with other – the input layer and then hidden layers and then the output layer. Dataset is passed as input and the input layer passes information / data to the hidden layers and output of each hidden layer becomes input of the next layer until the last layer is reached. As information is passed through the layers the machine learns it similar to human brain.

Figure 3- Human brain neural network vs. Artificial intelligence neural network

Steps to transform IT operations with Artificial intelligence

The approach proposed in Gartner article is the 12 steps approach. We start in the establishment phase and then move to the reactive phase, a process of inventorying existing skills and assets and then analyzing this information. This creates a solid platform to move on to the proactive and expansion phases, and begin truly transforming IT operations.”

Figure 4- Steps to transform IT operations to Artificial intelligence

Figure 5- Strategic planning to implement AI machine

List of few Artificial Intelligence software

Below is list of few AI software

  • Astro – Modern email and calendar apps for Mac, iOS, Android, Slack and Amazon Alexa
  • OpenNN is a software library which implements neural networks, a main area of machine learning research.
  • H2O– Open source machine learning platform for predictive big data analysis, data scoring, and data modelling
  • Braina– Multi-language speech recognition software with the ability to dictate in any third party software or to fill forms on websites
  • nanoRep– nanorep is a digital self-service solution that provides personalized guidance to every consumer at their moment of need, on any device
  • Recast.AI– Recast.AI allows developers to easily create bots and companies to improve customer support with automated agents
  • Amy – Meeting scheduling software with AI capabilities that emails with appropriate people and creates meeting invites.
  • Clarifai– I mage recognition API for developers
  • Lumen5– Video creation platform driven by A.I. and designed for businesses to produce social videos for content marketing.

Technologies and frameworks

Below are the generally used technology to create an artificial intelligence machine

  • R
  • Python
  • Spark
  • Julia
  • Mahout

Spread the word. Share this post!

Leave Comment

Your email address will not be published. Required fields are marked *