What is Artificial Intelligence

For the last five years, I have been working on older versions of Oracle Forms and Reports. Plus, I work with a government client. The above factors together have made my life peaceful and monotonous. So, one day I thought,बहुत हो गया अब पैसा नहीं डॉलर कमाना है.

So I started googling to explore which technology will narrow down the wealth gap between Jeff Bezos and Me. It took me about a week to settle on learning AI(Artificial intelligence). Back then, I thought It’s a matter of months until I will start coding my robot and Alexa.

Initially, it was like standing at a crossroad, getting confused and scared, then someone suggested starting with Python. I obliged and started reading Python. Days passed, and my dream of coding my robot was like a mirage in the desert. It was only after completing data visualization and machine learning by Python I got some clarity around AI.

AI Application around us

Let us see few AI application around us which we see but still don’t see.
Alexa:- It’s millions of codes analyzing what you are asking and replying accordingly.
Friend suggestion in social media:-no, it’s not Big boss. It’s an intelligent system learning from your friend list, familiar friends, recent friend, etc.
Movie suggestion in OTT platform:-Movie suggestion is based on your taste, your recent watch.
Frequently bought together option in e-commerce site:-codes analyses shopping patterns and recommends combination based on that pattern.
All this happens with minimal or no human intervention, and this concept is AI, Creating a system that can generate intelligence.

Understanding Artificial Intelligence

Following points are gist of my understanding about AI.

1>AI is not some technology; it’s a concept that encapsulates many programming languages, algorithms(algorithms are steps to achieve some task), and technologies.

2>AI is mainly Data and Algorithm; You collect data, apply the algorithm to data and make an intelligent decision.

3>Programming language and tools to collect data and apply algorithms on it to generate an intelligent decision.

Components of Artificial Intelligence

As already explained AI is a superset of many technologies and concepts. Generally, it has the following components.

Data Science

Data science is about generating knowledge from data. It’s not as simple as it sounds. To develop an understanding of data, you need to collect data, remove irrelevant data(cleansing), fill in missing values, represent data in the correct format, etc.
For example, the following data represent a crime against India’s women (Data provided by NCRB). The present scenario is not difficult to analyze, but understanding from raw data becomes cumbersome when data is vast with many dimensions. Different technologies help us represent data in an easy-to-understand format like the graph, finding a hidden pattern, performing operations on a massive chunk of data, etc.
A few such languages which help us to do that are Python, Hadoop, R, etc.

I have used the Matplotlib library of python to generate the below graph, which gives more understanding and clarity about the data mentioned above.

Machine Learning

Machine learning and deep learning are about self-learning from data and then applying that learning without human intervention.

Think of teaching your daughter to identify a cat. You show her a white cat, and then you tell her it’s a cat. You give her both input(cat picture or a cat) and output(confirming it’s a cat). Next time you show her a black cat(information/input) and tell her it’s a cat(output).

In this whole exercise, you are giving both input and output, and what is your daughter doing? She is creating attributes of a cat in her mind, learning pattern?

Next time you show her a cat(input) and ask her what’s this, she will respond(output) it’s a cat.

What you did while teaching your daughter is called a learning model in machine learning. You provide the system with both input and output to help it look for patterns in data and make better decisions in the future based on the input and output that we provide. Training systems with more data will result in more accuracy in predicting the outcome.

After sufficient training, The system can provide output for any new input. The outputs are compared with the intended output. The learning algorithm /model can be changed if there is a massive gap between intended output and system output.

The following software can be used for machine learning. Underlying language may differ. For example, TensorFlow uses either C++ or Python.

1. TensorFlow.
2. Keras.
3. Scikit-learn.
4. Microsoft Cognitive Toolkit.
5. Theano.
6. Caffe.
7. Torch.

Deep learning

It is a subset of Machine learning.

It is a more advanced form of machine learning.
In ML, when output is not in line with desired output programmer changes the learning model whereas in deep learning, the system resolves its fault through its own built-in neural network(just like the brain).

In a Deep Learning system, there are many algorithms layer one above another. These algorithms continuously analyze data to conclude similar to a human mind.

Have you seen the latest application of Artificial Intelligence?

Also Read:-Best Books To Read