Jargon Busting

This article is a short summary of the video by One Fourth Labs

With all buzz going around in the field of AI , "AI in Automobile", "AI in Banking", the fine line between the jargon is getting misty.

Is Machine Learning and/or Deep Learning the same as AI? Is Data Science the same as Machine Learning? these questions are bound to arise for an individual new to the field of AI.

You might have come across these different terms ( or not ), but these terms hints towards a difference between the jargon.

Lets start with some simple definitions :

Artificial Intelligence : is the simulation of human intelligence processes by machines, especially computer systems. These processes include learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions) and self-correction.

Machine Learning : is the field of computer science that uses statistical techniques to give computer systems the ability "learn" with data, and without being explicitly programmed.

Data Science : is an interdisciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and insights from data in various forms, both structured and unstructured,similar to data mining.

Pattern recognition: is the automated recognition of patterns and regularities in data.

Computer vision: is an interdisciplinary scientific field that deals with how computers can be made to gain high-level understanding from digital images or videos. From the perspective of engineering, it seeks to automate tasks that the human visual system can do.

Deep Learning : is a subset of Machine Learning, and an artificial intelligence function that imitates the workings of the human brain in processing data and creating patterns for use in decision making.

It is important to understand that AI is a collective implementation of various practices and abilities, which correlate to provide intelligence

These definitions still do not completely clear the confusion. In order to understand the finer differences ,let's pinpoint of the various characteristics abilities of an intelligent being that a machine can imitate and fill up the table mentioned below, and further understand the components we require( from the jargon itself as well) for AI framework.

As humans we can

1. Identify things (related to vision )
2. Create and listen to sounds
3. Read and write
4. Perform decision making

Now we can start fitting our jargon related to these characteristics

As you can see the jargon is itself being fitted into our model of AI. These are the abilities we want our artificial intelligence to posses.

Now that we have selected the abilities, they allow us to do some corresponding tasks.

With Computer vision we can identify entities and differentiate between them( object detection), like differentiate between a picture of horse and a giraffe, identify hand written digits etc.

With Speech we can convert text to acceptable audio or vice versa (Text to speech ), and we can identify individual speakers among group of speakers, etc.

With Natural Language Processing we can differentiate documents with respect to their subjects, or create an automated answering system, etc.

With Planning/Decision Making we can create an automated driving machine( it would require input from other mentioned abilities like computer vision)

Now we move towards the methods we can implement to achieve these abilities.

The earlier methods knows as Expert Systems used can be boiled down to a bunch of if else statements that could differentiate on the basis of attributes provided to match the statements.This as you can see, was a very ineffective and redundant method.

A better solution to our problem came as the form of Machine Learning, where the machine started learning on the basis of the data provided to it.

Over time a family of methods knows as Deep Learning was introduced , which performed better than the earlier mentioned methods.

The last is the Reinforcement Learning that allows us to perform the decision making tasks in a unique way.

Now we have moved towards the bigger picture.

AI encompasses all of this

We will state something as Artificial Intelligence if it is able to perform some or all ( there can more number of abilities than which we have mentioned ) of the abilities to achieve corresponding tasks with the help of various suitable methods.

Now we can mark the jargon we have discussed accordingly to the class

Let's move towards pattern recognition which we haven't discussed till now.

All the abilities we discussed , itself involve some form of pattern recognition. Identifying patterns that help us to differentiate between a horse and a giraffe , identifying patterns in a language. Thus pattern recognition manifests itself as a part of the numerous tasks we are performing to imitate the abilities.

Computer Vision involves object detection, but provided with this image of a car at night time, it becomes difficult to apply computer vision methods.

This is where image processing comes in, transforming the night time image to day time and thus applying computer vision methods , significantly improves the performance

As to how the image processing systems achieves the task, we can use a particular subset of the machine learning algorithms we use in computer vision. A machine learning algorithm, fed with different images of day and night time could estimate parameters and with help of functions covert the image as required.

The last that is left in our list of jargon is Data Science

To understand data science , we first consider the definition of science itself. Here we conducted the experiment of collecting the data, and then representing it in a suitable manner, to derive conclusions( this is a part of descriptive statistics )

Now we can apply machine learning algorithms to the data we collected with the help of data science medium.

THANK YOU FOR READING THIS ARTICLE